What do these do?

[jason1025]

I simply dont truly now what these do. I have never seen a comparison with a control

 

 

 

 

Ca someone create an illustration showing the difference in these settings? Not the difference from each other but the difference in values from within themselves.

 

I cant find anything on either of these.

 

 

 

If the comparison illustration could be in this fashion or something like this that would be great

 

[robcat2075]

We got into soft reflection a while ago

 

posts 18-24:

 

http://www.hash.com/forums/index.php?s=&am...st&p=322831

 

Also TechRef p.234

[jason1025]

OK Thanks but what does soften on do under multipass?

 

Anyone know?

[paradymx]

OK Thanks but what does soften on do under multipass?

 

Anyone know?

 

Not sure how to describe it. It blurs the image some, the render isn't as sharp. For the longest time when I first started using AM I couldn't understand why my MultiPass renders were always blury til I discovered that option.

[robcat2075]

soften discussed:

 

http://www.hash.com/forums/index.php?s=&am...st&p=165102

 

I find the effect of Soften to be minute and unnecessary for any purpose I've had. Perhaps someone can demonstrate a useful scenario.

 

You can see the difference by rendering (multipass must be >4) a frame ON and a frame OFF, then dropping both as rotoscopes into a new A:M chor.

 

Zoom WAAAAYYYY in to a sharp edge, then flick the lower rotoscope visibility on and off to quickly compare the two.

[jason1025]

Interesting. I have always turned it on because I thought it helped with aliasing. I will have to do some tests.

 

I remember viewing a thread that outlined which render options gave no additional render hit once other options were turned on. Can someone help me find that thread?

[Gerry]

Hey Jason, that illustration about the zero depth for displacement maps was a big help. I was having a heck of a time just the other day "cheating" a zero depth because I assumed it would be 128. It ain't!

[jason1025]

Does anyone what the render hit is for soften on? I would test it myself but all systems are rendering.

[robcat2075]

If your middle grays aren't coming out middle gray, something is set wrong in Photoshop.

 

When I make a map in Photoshop my middle gray comes out middle gray and creates flat displacement like it ought to..

 

 

 

 

This 16 bit map was saved in EXR format. My Photoshop doesn't do 16 bit PNG

[jason1025]

Hey Jason, that illustration about the zero depth for displacement maps was a big help. I was having a heck of a time just the other day "cheating" a zero depth because I assumed it would be 128. It ain't!

 

 

 

Here are some nice visual compares

 

AM_Compare_Notes.pdf

[zandoriastudios]

If your middle grays aren't coming out middle gray, something is set wrong in Photoshop.

 

When I make a map in Photoshop my middle gray comes out middle gray and creates flat displacement like it ought to..

 

 

 

 

This 16 bit map was saved in EXR format. My Photoshop doesn't do 16 bit PNG

 

I see that Photoshop gives me 149 as the number for 50%gray 0 for 100%black and 255 for 100%white. But why is that the case??? ( I always thought that 128 was the middle value)

[robcat2075]

I see that Photoshop gives me 149 as the number for 50%gray 0 for 100%black and 255 for 100%white. But why is that the case??? ( I always thought that 128 was the middle value)

 

 

When I set my gray to 127 and paint that on a map, as I did above, I get a map that does what middle gray is supposed to do. Zero displacement.

 

Someone's got a gamma setting somewhere not set to neutral OR maybe they have Photoshop's "Color Management" turned on. Turn that off so that numerical color values retain their numerical color values.

 

I have yet to read a cogent explanation of of how get Color Management to not mess things up for CG work.

[zandoriastudios]

that WAS what the numbers gave me! Are you positive that your surface isn't being depressed slightly? I am doing a test with a 149 and 128 dot on a transparent .png, should see NO displacement (one one of them)...I will edit this post in a moment

 

[Edit]

Here is the dots shown as a color map

 

 

128 As Displacement

 

149 As Displacement

 

 

It seems to me that either the 128 or the 140 should have shown no displacement...But obviously that is not what is happening. Thoughts?

[robcat2075]

Here's my EXR depth map

 

EXRGrayZip.zip

 

 

 

Open EXR plugin for Photoshop

 

 

Open it and see if the gray is indeed 127?

[John Bigboote]

HEY! Thats MY old illustration...I remember doing that as a test and then having Yves 'chew me out' for some reason or another...good times.

[jason1025]

I take no credit for these illustrations. I just put them up again.

[robcat2075]

that WAS what the numbers gave me! Are you positive that your surface isn't being depressed slightly? I am doing a test with a 149 and 128 dot on a transparent .png, should see NO displacement (one one of them)...I will edit this post in a moment

 

I'm not sure a transparent PNG is a reliable tool here.

 

A displacement map is a grayscale map. Once you start adding transparency to gray...

[zandoriastudios]

the transparency is to help me see the undisplaced surface, next to the gray spot...

[zandoriastudios]

Here's my EXR depth map

 

EXRGrayZip.zip

 

 

 

Open EXR plugin for Photoshop

 

 

Open it and see if the gray is indeed 127?

yep, 127

 

I tried another experiment, this time doing black and white and a series of grays between 126-129 (still using an alpha, so I can see the undisplaced surface), this time a targa:

But still the middle squares displace the surface.... I don't understand

 

I tried Robert's EXR, but to save with an alpha I had to reduce the bit depth...you can see that the 127 gray area still displaced the surface. but the 32 bit EXR behaved properly (but the color 127 is unchanged)!

[robcat2075]

Gray 127 made a slight displacement when I cranked the level up high. 128 seems to be true middle.

 

Here's a test with gray=128 and displacement set to 10000%

 

There's no displacement created where the gray decal meets the undecaled portion of the model.

 

 

 

 

 

I tried another experiment, this time doing black and white and a series of grays between 126-129 (still using an alpha, so I can see the undisplaced surface),

[zandoriastudios]

Found a different .EXR plugin that would let me make an alpha in the .exr:

[mtpeak2]

This may be a stupid question, but why not just use a black and a white dot with the rest of it an alpha? Why use the gray at all in your map?

[NancyGormezano]

In my observation, it seems that displacement behaves differently depending on how the pattern aligns with the splines - I used the same image for displacement and color (to show how it aligns), using gray values of 0, 128, 255 (tga), applied to flat grid, but did not cover the whole grid. 4 pass no soften. I haven't tried using exr images for decal.

[HomeSlice]

Sorry for hijacking your thread Jason ....

 

I did the displacement tests too. All renders are 3-Pass. Displacement is set to 10,000%. My middle gray in all tests is (128,128,128). I used an 8bit/channel TGA decal with and without an alpha channel, a 16bit/channel PNG decal with and without an alpha channel, and a 32bit/channel exr decal with and without an alpha channel.

 

I'm not sure if "alpha channel" would be the correct term for the PNG and EXR images. In BOTH cases (in Photoshop CS3 anyway) , if you want transparency, you have to delete part of the image (so the default checkered background is visible) because you cannot save a proper alpha channel. But in A:M, the transparent areas work just like an alpha channel would.

 

In every case, there was no displacement in the neutral area when there was no alpha channel.

But where an alpha channel was present, the neutral color WAS displaced.

 

8bits tga decal

 

8bits tga decal with alpha channel

 

16bits png decal

 

16bits png decal with alpha channel

 

32bits exr decal

 

32bits exr decal with alpha

[GruberR]

A side note to transparent PNGs from Photoshop: I have found out (through programs that don't support transparent PNGs) that the color channel usually contains garbage in the "fully transparent areas". This isn't an issue normally (because these are invisible anyway) but maybe the displacement map ignores the transparency information and uses the "garbage" to displace the surface? My 2c...

[robcat2075]

Is there any value you can put in the gray and not have the edge of the alpha make a visible ridge?

[HomeSlice]

Is there any value you can put in the gray and not have the edge of the alpha make a visible ridge?

The magic color for Displacement maps with Alpha channels appears to be black (0,0,0).

I used a TGA image for these tests.

 

128,128,128 background with a hard-edged alpha channel

 

128,128,128 background with blurred alpha channel

 

Black background with hard-edged alpha channel

 

Black background with blurred alpha channel

[robcat2075]

I can tell you that with all the displacement material tests I've done the 127-128 neighborhood gave me no displacement.

 

In those black background tests, the depression for the black spot is gone. that tells me that ,for displacement purposes, the balck alpha area is being interpreted as 0,0,0 gray resulting in full depression.

[HomeSlice]

I can tell you that with all the displacement material tests I've done the 127-128 neighborhood gave me no displacement.

As long as I don't have an alpha channel, my tests agree with yours.

 

the balck alpha area is being interpreted as 0,0,0 gray resulting in full depression.

The decals don't cover the whole model, so if this statement were true, the displaced area with the black would appear to be lower than the un-decaled edges. But they are all the same, which would lead me to the conclusion that when an alpha channel is present, black is interpreted as "Has no effect whatsoever".