Using the Alpha Channel

[Rodney]

How do I run the render again with the group images, and how do I place them?

 

Run (don't walk) to the nearest mouse click and view JohnL3D's tutorial on converting sequential TGAs to AVI and MOV.

Thats not all you can do with 'Save As Animation' but its an important start.

 

http://www.hash.com/forums/index.php?showtopic=20996

 

The direct link to John's WMV file is:

http://johnl.inform.net/images/convert.wmv

 

This 'Save As Animation' functionality replaced the old way of doing thing via camera rotoscope.

The downside of this outstanding (but apparently secret) timesaving feature? Knowledge of creating movies via camera rotoscope has been almost entirely lost!

 

Luckily for all of us camera rotoscopes are right along the path to understanding the power of Alpha Channels in A:M where we can get a great view of them too.

[Moonsire]

Run (don't walk) to the nearest mouse click and view JohnL3D's tutorial on converting sequential TGAs to AVI and MOV.

Thats not all you can do with 'Save As Animation' but its an important start.

 

The direct link to John's WMV file is:

http://johnl.inform.net/images/convert.wmv

 

This 'Save As Animation' functionality replaced the old way of doing thing via camera rotoscope.

 

Luckily for all of us camera rotoscopes are right along the path to understanding the power of Alpha Channels in A:M where we can get a great view of them too.

 

I can see how that will join my TGA files into a .mov, but, how do I do that <b>and</b> intagrate the single background image?

I tried to load the image sequence of 240 images, and it said there were too many. And once I have them there, the "Save as Animation..." feature does not come up for multiples.

[Rodney]

I can see how that will join my TGA files into a .mov, but, how do I do that and intagrate the single background image?

The simple answer is "You don't".

The 'Save As Animation' feature is for conversion... not editing.

 

There is an exception here (isn't there always?).

Using A:M's Composite feature you can build animation then Right Click and save out.

We'll have to investigate that later.

 

For editing (and our purposes) you'll need to use a Camera Rotoscope or Layers.

 

I tried to load the image sequence of 240 images, and it said there were too many. And once I have them there, the "Save as Animation..." feature does not come up for multiples.

 

Strange. I've can't recall ever having too many images in a sequence. I'm sure its possible but it has to be more than 240 images.

 

I don't follow you on the 'multiples'... not sure what that is.

 

This might be a good time to mention that you can break your project down into shots and compile them separately. The easiest way to do it would be to take a file (or files) in the sequence and either move it/them to another folder or rename. When A:M reads the sequence it'll stop when the sequence is interupted and only import those.

 

So... If you put 1-100 in one folder, 101 -200 in another etc you could compile those separately.

 

Note: I'm NOT suggesting this is something you want to do. Its an option that might work in some cases.

 

I'm curious about your 240 frame limitation. I need to investigate that.

[Rodney]

Update:

 

I imported and rendered out 300 sequential Targa images successfully.

They were low resolution as I mainly just wanted to make sure the number of images wasn't the problem. Could you be running out of memory?

 

 

If you get tired of waiting for my responses you can investigate the Composite feature yourself by Right Clicking on the Image container and experimenting there.

I'd probably be a good idea to download Noel Pickerings Tech Talk on Compositing to guide you through. It can be found in the Support Area, Tutorials forum pinned at the top of the forum. Look for 'Tech Talks'.

 

You'll really want to use images with Alpha Channels in the Compositor though if you are combining images in a scene.

[Moonsire]

 

O.K.

Let me restate what I should have had clear the first time.

 

1st. I have 030 to 278 in a TGA sequence, and A:M would not load them all at once. I did get them to load in groups of 100. I didn't try other groupings.

 

2nd. O.K. I get now that you don't use the "Save as animation" feature to create a .avi or .mov from TGA.

 

3rd. One uses the composite feature, which is talked about in Tech Talk #11 by Noel Pickering. (Once I know what to look for, I can often find it).

 

4th. You said "For editing (and our purposes) you'll need to use a Camera Rotoscope or Layers." Cool.

 

You are going to get a bit of a break from me. Tomarrow, and part of Wednesday, I will be out of town. My father died yesterday, and Wed. is the memorial service at my sisters. (Not too far from where I live). My brothers are coming into town also, so I'll be off line till Wed. evening.

 

Thanks for your help and time. I know I have taken up more than most, but I also have learned some things here that will be a big help to me in later projects.

 

Phil...

[Rodney]

1st. I have 030 to 278 in a TGA sequence, and A:M would not load them all at once. I did get them to load in groups of 100. I didn't try other groupings.

Very strange. Perhaps there was an image that broke the chain.

There appears to be a bug in here somewhere. I seem to have successfully converted a sequence of 1000 images to MOV format but now when I go back I can't import *any* sequential image. A restart of my system is probably in order.

 

2nd. O.K. I get now that you don't use the "Save as animation" feature to create a .avi or .mov from TGA.

 

But... um... you can. John's video tutorial should have demonstrated that.

Thats rather the whole reason to use "Save As Animation".

 

 

3rd. One uses the composite feature, which is talked about in Tech Talk #11 by Noel Pickering. (Once I know what to look for, I can often find it).

This is a great feature that I need to explore myself.

 

4th. You said "For editing (and our purposes) you'll need to use a Camera Rotoscope or Layers." Cool.

 

I'm reaquainting myself with Rotoscopes now. Its been awhile.

I prefer layers but sometimes the restrictions of rotoscopes help us keep things simple.

 

You are going to get a bit of a break from me. Tomarrow, and part of Wednesday, I will be out of town. My father died yesterday, and Wed. is the memorial service at my sisters. (Not too far from where I live). My brothers are coming into town also, so I'll be off line till Wed. evening.

I'm sorry to hear that. May God Bless you and your family and give you the strength you need for each other.

 

Thanks for your help and time. I know I have taken up more than most, but I also have learned some things here that will be a big help to me in later projects.

 

It is my pleasure. I could only hope that when you return I'll be more organized.

 

Take care!

[Rodney]

Correction:

 

Not really a bug. More of a screen refresh problem (probably due to memory as I'm running several programs at the same time).

 

While the PWS displayed as if there was only 1 image there were in fact 1000 frames imported. I imported again starting from 'image0002.tga' and it successfully imported the 999 frames up to 'image1000.tga'.

 

A Right Click and "Save As Animation' resulted in a MOV file comprised of the 999 frames. This process took about 2 minutes from start to finish.

 

To test my own theories I had renamed 'image1001.tga' as 'break_image1001_break.tga'. As predicted A:M stopped at 'image1000.tga'.

No real surprise there but nice to have the validation.

 

I have some more complex theories of compositing I'd like to try but one step at a time. Its important that we answer the questions at each level as much as possible before moving on.

 

Others are welcome to join in so that Phillip has some information to return to.

Any Alpha Channel, Compositing experts out there? Are we the only ones that care about Alpha Channels and sequential TGA conversions? Could be but I doubt it.

[Moonsire]

Very strange. Perhaps there was an image that broke the chain.

My sequnce starts at 0030. Would that be a problem?

 

 

But... um... you can. John's video tutorial should have demonstrated that.

Thats rather the whole reason to use "Save As Animation".

The tutorial shows the .mov file as a single file which he converts to a greater compression, and then takes the .mov file and converts it to a tga files. He does not show how to take the tga files and turn them into a .mov file.

 

When I load the tga files, they run from top to bottom, and no matter how I click on them, or the image file folder, the "Save as animation" option never becomes availible.

the files list like this:

1

2

3

4

5

6

Not like 1 2 3 4 5 6

 

I have tried everything I can think of.

 

 

I'm sorry to hear that. May God Bless you and your family and give you the strength you need for each other.

 

 

Thankyou.

[Caroline]

Are we the only ones that care about Alpha Channels and sequential TGA conversions?

 

Definately not! But you're going too fast - I shall have to catch up on all that I missed today.

 

And, Phil, you can do alphas with Photoshop Elements - only black and white ones though, I think - I've used Photoshop Elements to do simple alpha selection in the fly particle movie that I did. All you do is a selection, then you do Selection Menu > Save Selection, and save it to Alpha 1 (or whatever). You then save to a .tga, I think 32 bit, but I'm not sure that matters, I can't remember for the moment, and it's a working week (bother work).

 

I also have Paint Shop Pro 5 off a magazine, in which it is easy to check the alphas.

 

Photoshop one day maybe, but Elements has been good enough for me so far and is 1/5 the price, and being so used to it I much prefer the workflow.

[Moonsire]

Are we the only ones that care about Alpha Channels and sequential TGA conversions?

 

Definately not! But you're going too fast - I shall have to catch up on all that I missed today.

 

And, Phil, you can do alphas with Photoshop Elements - only black and white ones though, I think - I've used Photoshop Elements to do simple alpha selection in the fly particle movie that I did. All you do is a selection, then you do Selection Menu > Save Selection, and save it to Alpha 1 (or whatever). You then save to a .tga, I think 32 bit, but I'm not sure that matters, I can't remember for the moment, and it's a working week (bother work).

 

I also have Paint Shop Pro 5 off a magazine, in which it is easy to check the alphas.

 

Photoshop one day maybe, but Elements has been good enough for me so far and is 1/5 the price, and being so used to it I much prefer the workflow.

 

OK. Really. It's bed time. One more reply is all I have left before my wife drags me kicking and screaming from the computer.

 

Caroline: I have Elements 3, PhotoImpact Pro, and Gimp (newest). I like to use Elements and been doing tutorials. Gimp does color tga Alphas.

 

Now I'm off to bed.

 

Phil...

[Dhar]

Very late chiming in here, but that warehouse gave me trouble too when I was participating in TAoA:M exercises. It's the only set that took very long to render.

[Rodney]

Very late chiming in here, but that warehouse gave me trouble too when I was participating in TAoA:M exercises. It's the only set that took very long to render.

Dhar,

I'm not sure why that one would take excessive time to render.

There are a lot of variables in there.

 

If you still have the project file we could disect that as part of our exploration.

 

Definately not! But you're going too fast - I shall have to catch up on all that I missed today.

 

Caroline,

 

...and here I feared we were going so slow we might lose everyone's interest!

 

From time to time we'll post a recap (perhaps in the form of a PDF article?) to summarize what we think we've learned. Its important the information sinks as much as possible as once stuck we won't have to revisit the basics and can move on to more creative endeavors.

 

I find the process of exploration to be as much fun as the creative side but then I'm easily entertained.

 

There are so many related branches in this particular tree that the discussion might likely never end. Our goal here would be to at least satisify everyone's curiousity about Alpha Channels and hint at the ways we can use them.

[ypoissant]

I'm not sure why that one would take excessive time to render.

There are a lot of variables in there.

 

If you still have the project file we could disect that as part of our exploration.

And...

The building is the brick warehouse in the Model/buildings library in A:M. Almost everything is made up of magerials. Only a few little decals for dirt and grime. Maybe I need to pick my buildings better. But I don't have one with decals, and that means building my own. My own attempt at building brick walls has been less than spectacular, but I haven't given up yet.

 

Rodney, This model is from the CD and is full of materials. I ran two quck render tests. I droped the warehouse model in the standard choreography, repositioned the camera and rendered at 640x480. The render took 5m4s. Then I replaced all materials with simple attributes and the render took 0m20s.

 

Phil, IMO, your learning time would be better spent if you learned how to do decals and apply them. You don't need to modal the building. Just replace all the materials by decals on the existing model. This solution seems to be so much simpler than what you are currently doing.

[Moonsire]

Rodney, This model is from the CD and is full of materials. I ran two quck render tests. I droped the warehouse model in the standard choreography, repositioned the camera and rendered at 640x480. The render took 5m4s. Then I replaced all materials with simple attributes and the render took 0m20s.

 

Yves, When you say attributes, your talking about the items under the "Surface" heading? (color, roughness, transparence, etc...) I really like the brick look, but I am always going back to how much is that brick look worth in time and effort.

 

Phil, IMO, your learning time would be better spent if you learned how to do decals and apply them. You don't need to modal the building. Just replace all the materials by decals on the existing model. This solution seems to be so much simpler than what you are currently doing.

I have done decals and have a pretty good idea how to place them, though more work needs to be done on non-regular surfaces. I tried for a decal brick exterior and still have a way to go on getting one that I like. I tried a few brick tutorials and still didn't get a brick look I like. I like the idea of striping the building of its materials and trying a decal/attribute approach. It just might work.

 

Very late chiming in here, but that warehouse gave me trouble too when I was participating in TAoA:M exercises. It's the only set that took very long to render.

 

Dhar, I have the street set peice with the sidewalk, and the two of them together, with that little fire plug added up to 16 minutes. I think I'll look to using Yves suggestion in the future, and aim for more decals and attributes. But I am still wanting to work with Alpha Channel, as I can see much there that I may be able to use in the future.

 

Still, I also need to find out why I can't get the "save as animation" option to pop up when I want to convert the tga files to .mov.

 

I am off to my sisters for my dad's memorial, but I will likely borrow her computer to visit here.

 

Thanks all,

Phil...

[ypoissant]

Yves, When you say attributes, your talking about the items under the "Surface" heading? (color, roughness, transparence, etc...) I really like the brick look, but I am always going back to how much is that brick look worth in time and effort.

To me, at least, doing a brick decal looks like much less time and effort than doing this alpha composition.

 

I have done decals and have a pretty good idea how to place them, though more work needs to be done on non-regular surfaces. I tried for a decal brick exterior and still have a way to go on getting one that I like. I tried a few brick tutorials and still didn't get a brick look I like. I like the idea of striping the building of its materials and trying a decal/attribute approach. It just might work.

If you already like that brick texture, then you just need to render that wall from the modeling window in a high resolution targa file and then apply that back as a decal. You will need to do the same thing for the side of the columns too. Dong the bump maps should be easy once you have the color decals ready.

[Caroline]

I think I'm caught up with the situation now.

 

I can see that in this particular situation a decal renders much faster than the materials, however it is useful to have the skill to do something, even if not immediately required. (It's called "come-in-handy", like junk in the attic. )

 

I have my project set up with the wall, street and fire hydrant, just like the first picture. I have Thom walking down the street. I have default lighting, with an extra light to cast stronger shadows. (I'm not very good at lighting - I have put that on the "later" list.)

 

My projected workflow is:

1. Render one frame of just the wall, street and fire hydrant, in highest quality, 16x multipass, with shadows (reflections maybe, haven't understood them yet) to a tga.

2. Render just Thom walking down the street, with wall, street and fire hydrant not active, multi-pass 1, with alpha on, no shadows or reflections, to 3 seconds of tgas. Tgas will have the alpha of Thom.

3. Render just shadows of Thom walking down the street to 3 seconds of tgas. tgas will have the alpha of Thoms shadow.

 

1. to be achieved in about 50 mins (rendering now). (the lighting is horrible though)

2. achieved in 28 mins.

3. having difficulty rendering just shadows - am reading various threads now. I thought I'd set it up correctly, but I'm just getting blue sky with no alpha channel, so I've made an extra light that casts a definite shadow.

 

Am I on the right track to join in the discussion?

 

(There are so many aspects to this - I have learned only this morning about the shadow buffer. By exploring how to do this, I watched a composite tech talk - wow - much later. I also discovered Post Effects - also later.)

 

It will be really exciting to have a thread exploring a simple scene like this and how you can put it together in various ways.

[Caroline]

On the assumption that I'm talking about the same thing as you're talking about, I've been having some fun.

 

1. Render of main street with 16x multipass, shadows on - 52 mins.

Compressed jpgs, so quality is not as good as original.

 

2. Render of Thom into alpha channels - I made all other objects ACTIVE = OFF, and rendered with Alpha Buffer = ON, no shadows. - Render 28 mins.

(jpg here, so no alpha channel, but there is in my tgas)

 

3. Render of shadows - finally worked out settings. Under Options for warehouse and street, Receive Shadows AND Shadows Only have to be ON. I set Fire Hydrant to Cast Shadows = OFF, but this only saved 2 secs per frame. The render options: Alpha Buffer = ON and Shadow Buffer = ON. I did Multipass = 1, and Shadows =ON.

Total rendering time of 91 frames (3 seconds) was 2 hours 10 mins. (I think - I forgot to check).

Pretty scary - 2 hours of rendering pure black frames. The alpha channel screen capture from PSP looks like:

 

4. New project in which I imported the main street tga, the tga sequence of Thom, and the tga sequence of shadows.

New choreography. Deleted ground. Set up 3 rotoscopes on the camera - first main street, then shadows then Thom.

44 seconds to render, with all settings off, no multipass.

 

Total rendering time: 3 hours 30 mins, 44 secs.

To render original would be approx 89 hours (mostly just due to materials).

 

This is the movie result:

ThomSidewalk.mov

 

In retrospect, Thom walks in front of the fire hydrant (doh!), so the firehydrant would need to be rendered at the same time as Thom, and firehydrant shadows with step 3 shadows.

And in the original, Thom's walk action sinks into the sidewalk, so it looks a bit odd in the composite where he's a layer on top of it.

 

Rendered a still of frame 1:00 - 58 mins.

(Sorry, the lighting sucks )

and the composite render of 1 second

 

 

I do hope we're talking about the same thing.

 

Next is to work out how to put reflections in - I was thinking of just rendering the windows, but will have to sleep on it.

[nimblepix]

Caroline,

 

Think about it this way.

If Thom was walking in a real world shot,

you would make objects that receive the camera map

of the real world scene and catch shadows.

Do the same with the rendered background scene here.

The objects are already perfectly in place and ready to receive the map.

Thom can walk behind objects easily in a case like this

as long as those objects are in the same shot when he does the walking.

[Rodney]

Nice start Caroline!

 

As art direction comes down and it is determined new elements need to be added to the scene it can be adjusted too. As has been mentioned shadows and reflections need to be considered.

 

With A:M's ability to change models under the shortcut in a choreography you could change characters, add new elements passing in the foreground, add new windows and doors, graffiti, titles... and have then all animated too.

 

As Yves has mentioned it'd be best in most cases to have all the elements determined and arranged beforehand but sometimes that might not be possible. If someone is working on the background models for instance (as its more complicated and will take longer) a proxy model can be substituted in the interim.

 

Another place where this process is useful is in combining Live Action with CG.

A combination of imagery with Alpha Channels and Front Projection Targets would be ideal and allow maximum flexibility in post production.

 

Imagine an ad where Thom is walking past various monuments or the Great Wonders of the World. If planned right the background and even foreground elements could be swapped out easily while Thom just keeps on walkin' along.

 

Another benefit?

You can then take all these various elements and use them in other programs (Photoshop, CorelDraw, the list is endless...) to create derivative works.

 

In retrospect, Thom walks in front of the fire hydrant (doh!), so the firehydrant would need to be rendered at the same time as Thom, and firehydrant shadows with step 3 shadows.

 

Or if necessary it could be rendered separately and placed in the foreground.

A complication to this is whether shadows from any object might fall on Thom as he walked behind it. If that'd be the case then it'd be best to render then in the same pass.

[Moonsire]

On the assumption that I'm talking about the same thing as you're talking about, I've been having some fun.

 

1. Render of main street with 16x multipass, shadows on - 52 mins.

 

2. Render of Thom into alpha channels - Render 28 mins.

 

3. Render of shadows - Total rendering time of 91 frames (3 seconds) was 2 hours 10 mins.

 

4. New project in which I imported the main street tga, the tga sequence of Thom, and the tga sequence of shadows.

New choreography. Deleted ground. Set up 3 rotoscopes on the camera - first main street, then shadows then Thom.

44 seconds to render, with all settings off, no multipass.

 

Total rendering time: 3 hours 30 mins, 44 secs.

To render original would be approx 89 hours (mostly just due to materials).

 

DY-NO-MITE! This is what I'm talking about. Well done Caroline. I will give this a try on Thursday when I get home from work. That is exactly the type of approach I am looking for on doing some of these scenes.

 

In retrospect, Thom walks in front of the fire hydrant (doh!),

That's an easy fix. Top rate job Caroline.

 

Next is to work out how to put reflections in - I was thinking of just rendering the windows, but will have to sleep on it.

 

 

I'm sure there will be an answer to this soon enough. We could just make duplicates of the windows and put them just over the background windows.

 

I am going to filddle with this for a little bit before I have to go to sleep. I'm not do on set until 11:30 am tomarrow (doing "24"), so I get to sleep in some.

 

Phil...

[Moonsire]

Once again, I want to ask if anyone can think of a reason why I can not load more that 100 images (TGA) at a time.

 

And when I watched the tutorial, the images were stacked left to right, is this really the way they are supposed to be?

 

And why can't I load more that one at a time into the camera? I am assuming I am doing something wrong.

 

Phil...

[Caroline]

Thanks, Phil, it was really fun.

Don't know about over 100 photos - I can't 'play' until tomorrow

 

Which tutorial?

[Moonsire]

Thanks, Phil, it was really fun.

Don't know about over 100 photos - I can't 'play' until tomorrow

 

Which tutorial?

 

Rodney sent me to this tutorial to learn to convert the tga's to .mov, Only it doesn't actually show that specific funtion, so I don't know how the tga files are supposed to look in the images folder on A:M. Alas, so far I have only been able to load one tga into rotoscope/camera at a time.

 

The tut info is below. Very good tut for convert of .avi/.mov movies to smaller files or to alternate files, includeing TGA's.

 

The direct link to John's WMV file is:

http://johnl.inform.net/images/convert.wmv

 

I should get home around 8:00 pm tonight, and I will try to run the shadow render. At least, by the time I get the TGA to .mov figured out, I should have all the tga files ready to go.

 

later...

[Rodney]

Once again, I want to ask if anyone can think of a reason why I can not load more that 100 images (TGA) at a time.

 

Sure can.

 

If you render your files out to a filenname as follows:

 

image.tga

 

A:M will number the renders thusly:

 

image1.tga

image2.tga

image3.tga

image4.tga

image5.tga

image6.tga

image7.tga

image8.tga

image9.tga

image10.tga

...

 

Notice any difference in pattern in any of those numbers?

 

Right. image10.tga is going to be problematic.

Computationally, images will be sequenced thusly:

 

image1.tga

image10.tga

image2.tga

...

 

 

The way to avoid this is to pad the filename with trailing zeros. Like this:

 

image00.tga

 

A:M then will render the first and tenth frames thusly:

 

image01.tga

image10.tga

 

So then what about three digit numericals?

 

image09.tga

image99.tga

image100.tga

 

We experience the old padding problem again, right?

 

I've gotten in the habit of always adding three zeros (minimum) to most sequential renders even though I usually don't need that many. Its just habit.

 

sequence000.tga

 

This three zero padding will allow for 1000 images so may not be enough for larger compilations but will more than satisfy most rendering needs.

(Note: This will accommodate filenames from 000 to 999 so 000 is your 1st frame)

 

So try padding your filename with zeros and see what you can see.

[Rodney]

I should say also that once you've rendered your images out is an unfortunate time to realize the sequence doesn't work. All is not lost however.

 

There are utilities (both Mac and PC) that can be used to rename large numbers of images rather quickly. I like Irfanview (XNView is a similar program). These are pretty extensive graphics conversion freeware utilities but there are utilities that do nothing but rename too.

[Caroline]

Phil, I think I may know what you're doing.

 

I rendered out 180 tgas, then loaded them back into a new project, put them as a rotoscope to a camera, rendered out to mov all 6 seconds no problem at all.

 

When you are importing them, don't import them all.

 

Right click Images in the PWS, then click Import > Animation Or Image Sequence. Then ONLY click the first tga - like Shaggy0000.tga. Then click OK, and the others will all be there. It may only look like one frame, but they are all there. You can then apply this one to the camera rotoscope in the chor, and each frame in the chor will go through the tgas.

 

This is how it looks in my PWS on a blank project having imported 180 tgas.

 

But like Rodney says, the 000 sequence when you render is important, took me ages to work that out, , I really got to know renumbering in Irfanview.

[Moonsire]

Phil, I think I may know what you're doing.

 

I rendered out 180 tgas, then loaded them back into a new project, put them as a rotoscope to a camera, rendered out to mov all 6 seconds no problem at all.

 

When you are importing them, don't import them all.

 

Right click Images in the PWS, then click Import > Animation Or Image Sequence. Then ONLY click the first tga - like Shaggy0000.tga. Then click OK, and the others will all be there. It may only look like one frame, but they are all there. You can then apply this one to the camera rotoscope in the chor, and each frame in the chor will go through the tgas.

 

This is how it looks in my PWS on a blank project having imported 180 tgas.

 

But like Rodney says, the 000 sequence when you render is important, took me ages to work that out, , I really got to know renumbering in Irfanview.

 

Thank you Caroline. That was the missing peice of information. I was highlighting all the images 000 to 249. I just tried it and it worked fine. Alas, in most programs, if you want to import a group of images, you import them all by highlighting them all. Now I know better. When this is all done, we need to put together a simple step by step.

 

I have a question for you. I tried rendering my shadows (test rendering a single image). It took ten minutes to render, taking almost all of that time on rendering the invisible building. What am I doing wrong that I am getting a ten minute render time? I thought I followed your settings, and nothing showed in the viewing window or the rendered TGA (except the shadow of course).

[Caroline]

Great - I remember doing that too, because that's how other programs work.

 

About the shadows, I have a feeling that happened to me the first time, but because I was experimenting, I didn't know what to expect. Try exiting A:M and starting again.

 

I just checked again, and all I did was change those settings - make sure the shadow only for building & street is ON - make sure you've set that on frame zero, because I think it's animatable, so if you set it on frame 30, frames 0 - 29 won't accept shadows (?not tested?).

 

When you put the shadow only to on, deselect the building, and it goes away, and you just see blue sky. Same for street, so Thom looks like he's in the sky. Does that happen for you?

 

When you say nothing showed up in the viewing window, I'm not sure what you mean. My tga was entirely black, and I couldn't tell whether it had worked until I looked at the alpha channel.

 

To compare, because I may have a faster system (mine's dual 2.8Ghz, 1GB Ram), with those settings - multipass = 1, shadows ON, everything else OFF, my shadow took 1:32 Mins. Rendering the whole scene, multipass = 1, shadows ON, everything else OFF, the scene took (bother I just closed it) I think it was 3:30 - 4:00.

 

So from that I guess we can tell that shadow only isn't rendering the materials.

[Moonsire]

Great - I remember doing that too, because that's how other programs work.

 

About the shadows, I have a feeling that happened to me the first time, but because I was experimenting, I didn't know what to expect. Try exiting A:M and starting again.

 

 

I'll try following these instructions, though I need to do it tomarrow. My call is 8:00 am, and that means a 5:45 wake-up.

 

I managed to get a first render of the scene with building/background and characters (van, horn, and fire plug). Now, all I need are the shadows, and having seen the scene play out, I may need to fuss with the actions, to get the look a little more spot on.

 

I'll let you know tomarrow if this all works out.

 

Thanks.

Phil...

[Rodney]

Phil/Caroline,

 

Its been great seeing you both progress through experimentation.

There is no better way to have the information sink in that to dive in and do it.

 

I'm looking forward to your next update.

 

Have fun at work! (I need a vacation)

[Moonsire]

Phil/Caroline,

 

Its been great seeing you both progress through experimentation.

There is no better way to have the information sink in that to dive in and do it.

 

I'm looking forward to your next update.

 

Have fun at work! (I need a vacation)

 

Funny, my wife says the same thing all the time.

 

Hey, I tried it out, and guess what. It worked. I used the background, and the action TGA's and ended up with a .mov. Now that I've done this, I am ready to go back and rework the scene to get better actions out of my main character, and to use the low rez version of the building (which will be done by ridding it of materials) to make the shadows. Would I be right in the notion that I could get the reflections if I set up the render in the same manner as the shadow render (except with render reflections)? Just a thought. It will take a while to do this, but when done, I will post it here with notations about what steps I took.

 

This has been very educational and a ride worth taking. Thanks all.

 

Phil...

[Caroline]

Does the low-res version make the shadows quicker? Surely, it shouldn't make much/any difference.

 

No, there's no reflection render option. I've been pondering on this, reading This Post, which suggests doing a greenscreen.

 

I'm trying to work out how to do the reflections with an alpha greenscreen, though. I haven't got very far with that. Any ideas?

 

Of course, now we're almost getting into the realms of spending 150 hours rendering to openexr format and tweaking lighting in composite.

 

I still think it's useful to be able to render in parts like we're doing.

[Moonsire]

Does the low-res version make the shadows quicker? Surely, it shouldn't make much/any difference.

 

No, there's no reflection render option. I've been pondering on this, reading This Post, which suggests doing a greenscreen.

 

I'm trying to work out how to do the reflections with an alpha greenscreen, though. I haven't got very far with that. Any ideas?

 

Of course, now we're almost getting into the realms of spending 150 hours rendering to openexr format and tweaking lighting in composite.

 

I still think it's useful to be able to render in parts like we're doing.

 

Hey Hey, Ho Ho slower renders gotta go!

I stripped the building of all materials and decals, and ran the shadow render. When I ran it with the building as it was modeled, it took just over 10:00 minutes. See, Caroline, I don't have one of those fancy smancy gigafast beasts like you do. But, without all the extras, the shadow render took 13 seconds. Now that, I can live with.

 

As to greenscreen, I haven't a clue. This will work out great for me however, doing the composite work after quicker renders in parts. I'll play refections with your, but should we start a new topic for that or, continue it here? Rodney, any input there?

 

In the mean time, I know I've learned a lot about Alpha Channels in the last few days. Is there more, or "are we there yet"?

[Caroline]

My computer's even got a glass side, but as it's under the desk, only the dust bunnies get to see the eerie blue glow.

 

I think reflections here, because aren't we going to alpha channel those as well?

 

Did you catch this post from satyajit?

 

Can't see why it needs to worry about materials on a shadow render, though. Although perhaps shadows would be affected by bump and displacement maps, so perhaps it makes sense.

 

I'll have an hour tomorrow to play with a greenscreen.

[Moonsire]

My computer's even got a glass side, but as it's under the desk, only the dust bunnies get to see the eerie blue glow.

 

I think reflections here, because aren't we going to alpha channel those as well?

 

Did you catch this post from satyajit?

 

Can't see why it needs to worry about materials on a shadow render, though. Although perhaps shadows would be affected by bump and displacement maps, so perhaps it makes sense.

 

I'll have an hour tomorrow to play with a greenscreen.

 

Reflections here then. Satyajit had a good post.

 

The materials do effect the shadow it seems. Bumps and maps.

 

Let me know how the greenscreen works out.

[Rodney]

*A word of advice to newbies concerning green screens*

 

So interesting is the process of green screening that new users often can be found turning their backgrounds to green (or other color) in an attempt to replicate the 'real world' process. What they fail to realize is that they are adding an unnecessary step (several actually) into the process and complicating their work.

 

Within A:M its better to use Alpha Channels which in effect are green screens identified and processed for you.

 

Look at it this way, why add in a background that you intend to just take back out later?

Better to leave it out in the first place if you can.

 

Where green screens are used most effectively is in video footage taken from the real world.

Live action and even stop motion animation filmed on a stage or set can benefit from the use of green screens.

 

The screens (green, red whatever) are used to isolate objects from their background so that the objects can easily be extracted and used separately. Without isolating the background prior to filming the process of extracting elements would be much more difficult and laborious.

 

Is there no use for green screens in A:M then?

 

One possible use might be for when images created in other programs are being used.

In most cases it would be better to use the program it was created in to isolate the image but all programs are not created equally. Some just can't do the job.

 

The process of isolating elements with these *external images* would be to use one solid color not used in the elements desired to be retained and then use A:M's ability to identify a Key Color in an image as transparent. So if outlining a black and white image in red all the red could be made instantly transparent while everything black and white would be retained.

Its important to note that you want this color to be as uniform as possible. This means that lighting of the objects must be considered too.

 

With available masking tools in most graphics software this tedious process simply isn't practical anymore but can be used as an example of what time is being saved/lost in the process. If all images used in a production are required to be isolated in this way production becomes costly in terms of labor, processing time and in error checking and correction.

 

If all the indivicual elements of a planned composite originate in A:M then our task is easy.

Rendering elements out with Alpha Channels automagically isolates them for use in the final image.

 

So save green screens for live action footage and save yourself a lot of time.

 

I'm not suggesting this is new information to most who will read this but hopefully it'll save the occassional new user some time.

[Caroline]

Thanks, Rodney for the info on green screens.

 

The problem that I'm having with reflections, which may be insurmountable in the example we're using, is that I am unable to isolate only the windows into an alpha. When I delete the materials on the building, to decrease render time, the lighting changes, as you might expect, and the reflections are different. So it may be a question of rendering the whole thing very slowly on 1 multipass, then render the still on 16 multipass, then alpha the still in places, so that you can see through to the 1 multipass. But that seems way too complicated for what should be a simple compositing.

 

Still thinking.

[Rodney]

The problem that I'm having with reflections, which may be insurmountable in the example we're using, is that I am unable to isolate only the windows into an alpha. When I delete the materials on the building, to decrease render time, the lighting changes, as you might expect, and the reflections are different.

 

Hmmmm...

 

What happens if you just hide the building (in the model) and leave the windows?

[Caroline]

I think I'm coming to the conclusion that if you're going to use reflections, you must expect a long render time anyway.

 

I put up a greenscreen (red actually) over the building, except for the windows, and turned everything on the redscreen to OFF, and rendered reflections only = ON. That was 5 mins for a 3 second one-pass. For 16 multipass to match the original still tga, it'll take 1 hour 45 mins for 3 secs.

 

My reflections tga looks like:

 

and the resulting movie:

untitled.mov

 

Movie has these layers: Reflections (red layer), main street still with alpha channel where windows are (photoshop elements selection - I was hoping not to have to photoshop), shadows, thom.

 

I deliberately left red edges around the windows, where I wasn't very careful, so you could see the idea.

 

So this method of doing reflections separately, would only work where you are willing to compromise the quality of those reflections. Although, I guess, even if you did render for 1 hour 45, we're still only up to 5 hours 15 total render, as opposed to 89 hours of the original estimate.

 

Quiet note: you may realise that I am really bad at lighting and setting up reflections.

[Moonsire]

The problem that I'm having with reflections, which may be insurmountable in the example we're using, is that I am unable to isolate only the windows into an alpha. When I delete the materials on the building, to decrease render time, the lighting changes, as you might expect, and the reflections are different.

 

Hmmmm...

 

What happens if you just hide the building (in the model) and leave the windows?

 

Clever. Just might work. Then place this layer over the layer of the building.

 

I have rendered the background, and the character passes, and am rendering the shadows as I type. The grand total of render time should be less than 100 minutes (1 hour and 40 min). The I must composite them. And I might even try adding the reflections (or may just leave them out). All still less than 75 hours of work, much less render time.

 

We'll see how this works out by tomorrow.

[Moonsire]

Hi all.

 

Here is the first scene from Horn's Dilemma, thanks to the help of all of you. Rodney, I have learned a few things about doing the TGA renders with Alpha Channels and Shadows. You must make sure that the chor. is lined up at 00:00:00 or some of the things don't pop in till later. Example: Watch the attached movie and you'll notice the shadow on the fire plug (before and after the van). This is reduced from the original and thus a little less crisp, but should show you that I am learning from all of this.

 

Total Render time (total of 4 renders) for this scene is 1 hr and 58 min. Beats heck out of 75 Hours.

 

scene1.mov

 

Comments welcome

[Caroline]

Nice. You have reflections - how did you do them?

 

On the right hand window, the van's reflection shows for a long time. If it's supposed to be a moving van, not parked, then the reflection shouldn't be there for so long.

 

Cute character - I like his shininess (please change that word to some 3d technical term that I haven't learned to trot out yet ).

[Rodney]

Nice!

I see the shadow popping you mentioned.

One of the Cons of compositing versus rendering as a complete scene is that you have more elements to control independently. The Pro is that those elements can be adjusted independently.

 

*This is an important note for new users*

Learn to love frame 0. In many cases you'll want to be on frame 0 when adjusting things to keep them from being animated. If you make a mistake you may need to go back and adjust/delete frame 0 keys.

 

Two other related methods to keep from animating things you don't want to:

- Change the animation driver to 'Constant' (versus time based etc) for things that won't change in a scene

- Toggle on/off the Animate icon which will prevent keyframes from being created.

 

If you decided you wanted another car to pass in front instead of that one how long would it take you to implement that change? (A guess will do... I'm not suggesting you rerender to find out.)

[Moonsire]

Nice. You have reflections - how did you do them?

 

I took Rodney's suggestion, (I had already decided to try that anyway) and used model mode to hide everything but the windows. Then I left them in the scene with the character render. Hardly slowed things down at all.

 

 

On the right hand window, the van's reflection shows for a long time. If it's supposed to be a moving van, not parked, then the reflection shouldn't be there for so long.

I think part of the reason that Stalk (the model's name) turns is because he hears the van start up. I have noticed the Van in the window from day one, an felt like it belonged, assuming I could make all the reflections work.

 

Cute character - I like his shininess (please change that word to some 3d technical term that I haven't learned to trot out yet ).

 

You could say that you like his reflective nature, but shininess works for me.

 

Nice!

I see the shadow popping you mentioned.

One of the Cons of compositing versus rendering as a complete scene is that you have more elements to control independently. The Pro is that those elements can be adjusted independently.

 

The fix is as simple as my going into Gimp and copying the shadow and part of the fire plug, then doing a quick fix on the other images. I only need to do that for the first few seconds of the film. Or I can render those first few seconds again with the shadow.

 

*This is an important note for new users*

Learn to love frame 0. In many cases you'll want to be on frame 0 when adjusting things to keep them from being animated. If you make a mistake you may need to go back and adjust/delete frame 0 keys.

 

Two other related methods to keep from animating things you don't want to:

- Change the animation driver to 'Constant' (versus time based etc) for things that won't change in a scene

- Toggle on/off the Animate icon which will prevent keyframes from being created.

1. Where is the animation driver to change to 'constant'?

2. Where is the animate icon you speak of.

3. On of the problems I have had is that I make the character do something, and at the end, I change the way it looks or moves and that change is incorporated into the movement from the very beginning, starting when I put play. How to avoid that?

 

If you decided you wanted another car to pass in front instead of that one how long would it take you to implement that change? (A guess will do... I'm not suggesting you render to find out.)

 

I could do it quickly with the car.

Step one, import new car.

Step two, go to properties of old car.

Step three, click on Short cut to .... Mini Van and pick the name of the other car.

Job done.

I would have to render the scene again with the van, Stalk, fire plug and windows...total time:

 

Total Frames: 361

Rendered in Multipass: 2

Total Render Time: 00:48:43

Average Time per frame: 00:00:08

 

Actually a little less on render time, because I would only have to change the frames with the car or it's reflection. When it passed on the left, I would have all that I need.

 

If I wanted even more control, I could alway render each of the characters separately from each other. Then the plug fix would be even faster with only the plug to render.

[Rodney]

Thanks for the response.

 

Animate Mode... lets tackle that one first.

There's an image of what the Animate icon looks like at the top of the page here:

 

Tech Ref: Animate Mode

[Rodney]

There is minimal information in the Tech Ref on the 'Constant' driver as its pretty straightforward; changing a driver to Constant will imput one value for that parameter over the life of the animation. This is opposed to "Time Based" which allows us to increase or decrease values over the course of the action.

 

If you dig a little more information surfaces in the Tech Ref.

The following doesn't show up in simple searches:

 

Constant Drivers

By default, objects in Animation:Master automatically animate. If you move an object at a given frame, a keyframe is automatically created at that frame.

 

Occasionally this is not desirable; it can be particularly annoying with cameras and lights. For example, suppose that in a choreography, the main character does not walk on set until three seconds have elapsed. At that time you determine the lighting needs some changes. You want to move the lights, but you don’t want the lights to slide there starting at time 0:00:00.

 

This is exactly what animate mode is for. Select the light, deselect “Animate Mode” on the Properties Panel, then move the light. If the light will never be animated, just leave the “Animate Mode” off. You can set any light, camera, null, or model shortcut the same way.

 

Now you can see the results on the frame you care about, yet the key will be placed at time 0:00:00, and the results will be visible at the frame you are on. The preceding scenario is the most common reason to use this feature. However, the way this works when keyframes already exist is slightly more complex. Suppose that your light already had keyframes at 0:00:00, 0:02:00, 0:04:00 and 0:06:00, that formed a small circular motion. Now at 0:03:00 you wish this entire motion was keyed higher and farther to the left.

 

By turning off “Animate Mode”, you can make this adjustment at 0:03:00. Simply translate the light higher and to the left. No keyframe will be created at 0:03:00, but the light will follow your mouse. This is achieved by modifying all of the existing keyframes by the requested amount of change.

 

 

One easy way to see the Constant driver (so you know what to look for) would be to:

 

Create a New Project

Create a New Choreography

Right Click on the Camera and import a new Rotoscope

Right Click on the Image or Transparency under the Rotoscope

 

You'll see two options; Constant and Time Based.

 

Note that it may not work in all instances *if* you don't change it to Time Based and then back to Constant. You'll find there are a lot of settings where you'll want to perform this simple toggle on/toggle off in order to activate the option. Activation is important not only so you can use the driver (or keyframe) but also so that when using Filters A:M will recognize that it is activated and identify it via the filter.

 

Filters are related in a way to our exploration (especially for management and organziation of projects) so more on Filters later?

[Moonsire]

Filters are related in a way to our exploration (especially for management and organziation of projects) so more on Filters later?

 

Filters it is then.

[Rodney]

Phil,

I'm easily distracted so keep reignin' me in!

 

Here are some words of wisdom related to the rendering issues we've been discussing.

From the Tech Ref (everybody reads that right?):

 

RENDER SPEED ISSUES

 

A typical image rendered in Animation:Master is very fast, one of the fastest in the industry. However, it is easy to choke the renderer to a standstill. The common culprits are, (in order of frequency):

 

1) Raytracing operations like:

a. Mirrors

b. Raytrace shadows (especially multi-ray soft shadows)

c. Transparency from a “complex” material, or with “density” or “refraction” (but not plain transparency)

d. Boolean operations

 

2) Complicated “complex” materials (with many combiner nodes)

 

3) Roughness (because it requires a lot of antialiasing)

 

4) Hair! (especially for the A-buffer because thin geometry, like hair, has little inner-pixel coherence)

 

5) Image size

 

a. A 2000 X 2000 image takes 64 times longer to render than a 512 X 512 image. When using the A-buffer, images larger than 2000 X 2000 are never required.

 

6) Volumetrics (with small step sizes)

 

7) Complex light rigs

 

8) Radiosity (sometimes, a single image can take many, many hours – but acceptable radiosity times are possible)

 

9) Multi-pass (for 3 or more passes)

 

10) Particle systems

 

There is no recommendation of when to use particular render settings – art is subjective. Just be aware of the time penalties involved and set your expectations accordingly.

 

Next up: A word or two on Filters

[Moonsire]

Phil,

I'm easily distracted so keep reignin' me in!

 

Here are some words of wisdom related to the rendering issues we've been discussing.

From the Tech Ref (everybody reads that right?):

 

RENDER SPEED ISSUES

 

 

Next up: A word or two on Filters

 

I will look at these posts again on Tuesday. I had 5 hours sleep before I had to be on the set of "24". I am a little tired. I will post again, tomorrow.

Phil...

[Moonsire]

Phil,

I'm easily distracted so keep reignin' me in!

 

Here are some words of wisdom related to the rendering issues we've been discussing.

From the Tech Ref (everybody reads that right?):

 

Rodney, I work as a background actor, I read several books at the same time, usually at least one magazine, have several animation projects, a newsletter, webmaster 6 web sites, read dozens of blogs a day, work tutorials on photoshop elements and other similar software, research my family genealogy, read/correspond via email with family and friends, and try to keep up on my favorite t.v. shows. And you want me to reign you in?

 

 

RENDER SPEED ISSUES

 

1) Raytracing operations like:

a. Mirrors

 

Does this include reflections, like on my Stalk character?

 

b. Raytrace shadows (especially multi-ray soft shadows)

What is the other method for creating shadows?

 

 

2) Complicated “complex” materials (with many combiner nodes)

 

Learned my lesson on this one with that brick building. More materials than I have eye lashes.

 

3) Roughness (because it requires a lot of antialiasing)

Didn't even consider this one.

 

The Tech Ref: Animate Mode things didn't make much sense to me, but I'll look at them again. Then again, the Constant Drivers explination may have just put it into context for me. This is a problem I have had with other projects.