robcat2075

That original "sRGB" version is how every interior scene of the White household looked in "Breaking Bad" so it's not necessarily "wrong".

there is also something not-quite-right in his demonstration of "desaturation" I'm still looking into that... I should note that his lighting technique of lighting the entire room with the light bouncing off the wall... that is something the A:M renderer can only do with Radiosity which is very time consuming. I could light that scene to get a very similar result without radiosity but that is another issue.

It takes me a while to watch that because I have to pause it every three or four minutes to go vomit. Listening to him chuckle at his own jokes is maddening. From what i can see, "Filmic Blender" is a render to a high dynamic range format like OpenEXR with a drop down list of post processing presets that do what one would typically do in After Effects. Look at the image below full screen. On the top is his before and after comparison of "sRGB" and "Filmic Blender" On the bottom, I've applied a slight "Curve" adjustment and a "Level" adjustment to just the sRGB side. It's kind of like a "gamma" change" Is that result not very similar? Mine is very noisy because I just have a screen grab off YouTube and not the real render. But just that one change has gotten a very similar result. I'm not an expert but I'm not convinced that the Blender Guru is speaking expertly about all the things he is speaking about.

The basic problem is that A:M and any other 3D software can calculate the brightness of the Sun and total darkness and everything in between. However, common image formats like jpg or targa or don't have enough data range of numbers represent all that. They store values of 0 to 255. If 0 is black and 1 is the dimmest detail visible then you can only double brightness 8 times before you run out of numbers and start clipping values. However, that is also about all that the monitors we look at can show. About 8 doublings of brightness from top to bottom. Maybe 9 on great monitor maybe way less on a cheap one. Actual film formats, viewed in person, can do better. Print does worse. Example: If I take a picture of the Sun and then display that on my monitor or print it on paper, that image will not be as bright as the Sun was. So, for any display format, choices have to be made about what part of that nearly infinite range of brightness you could calculate is crammed into what it can actually displayed from min to max. OpenEXR renders let you store a far greater range of brightnesses... 30 doublings, I think... but you still have the problem that no display can reproduce that. In addition, there is the problem of color....

There are several statements in the first few minutes that I am extremely doubtful about such as the assertion that a CRT monitor has an inferior dynamic capability to an LCD monitor. Just that is provably wrong. His comparison of the photo and render at 6:09 completely misstates why those two look different. I haven't gotten to the end of it yet. His solution (have't gotten there yet) may work in a practical manner, but I sense that he doesn't really know what he's talking about. Again I haven't watched it to the end yet... but if more dynamic range is the key, that's what OpenEXR is for. That is ostensibly "linear data" untainted by any color management You can pretty much render to that and do curve adjustment in post to include or exclude any part of the dynamic range you wish. I watch some more later. I'll also note that Yves Poissant has made a number posts of about color management and A:M and that if you say his name three times he will appear.

I made a few tests your sample in v19, still using the same 0.15 thickness and Jitter set to 100% 1 pass 3 passes 16 passes 64 passes 256 passes

Cool-looking shots!

That is odd. I guess that's not where the ground plane ends. Maybe that's where the top of the sphere is?

I have sometimes wished there was a way reduce more than one channel at a time. Do you have an example case?

Hooray!

It works normally here... model window>select spline>>Plugins>Wizards>Sweeper

Someone is getting a head start on "Planes, Trains, Automobiles"

Just to try it, I rendered Tore's test project with no blur and added blur in post with After Effects "Directional Blur" I manually keyframed the size and direction of the blur for each frame. This took longer than either of the rendered blurs. testMB AEDirectBlur_.mov

I'm still on AE from the year 2000 but there are good DOF plugins for modern AE that can take the depth map in an OpenEXR render and create convincing DOF effects in post. Perhaps there is something like that for your compositing program. However, good motion blur is still pretty much about multi-pass, even in studio settings. I'll note that in a shot like this... ...DOF could be simulated by rendering the characters and background separately and selectively blurring some and not others.

Cautions about non-multipass motion blur... 1) It's basically a post effect filter that approximates motion blur. It works well in situations where an object is moving linearly against a static background. Rotational motion and moving the entire camera around a pivot point are more complex than it can do well. Supersampling (AKA Multipass) is there for those complex occassions and will give an accurate result. You do have to decide how many passes you need do to get a satisfactory blur. 2) Neither non-multipass motion blur nor non-multipass DOF work well against the default camera background color. They need a background that is solid patches. Here is the non-multipass Motion Blur effect in v19e. It does not have the strange result that you got, but still is less than ideal. Render time about two minutes testMB no MultiP BKG_000.mov Here is multipass motion blur with 16 passes. This seems more appropriate. Render time about 11 minutes testMB MultiP BKG_000.mov I tried non-multipass DOF with an old test PRJ and it does not work in v19e. That would be a bug. Are you using After Effects in your project? Or some other compositing app?

The blue specular color is very interesting! Where does that come from?

Mark, in Post #29 in this thread is a video that covers how to get A:M's EXR depth buffer into After Effects as we were wondering about at Live Answer Time today... https://www.hash.com/forums/index.php?showtopic=43473&p=377728

I like stylized heads. One suggestion... you can usually get a smoother appearing mesh going to the Surface Properties, turning Average Normals ON, and setting Normal Weight to 50%

This is the guy who gets everyone yawning at the weekly department meeting. Looks good!

It's basically a problem of momentum. The rider doesn't stop because the horse has stopped. The rider keeps moving after the horse stops and only stops because he falls and hits the ground.

Set the Pose ON in the Model before you make an Action for it or drop it in a Chor. I'm not sure what you are doing differently to get a different result. If you could bring this to Live Action Time we could look at it live.

Two problems... the Pose with the constraints needs to be on and then there is some hierarchy to adjust. clip4141MadFoxIKLeg2K.mp4 Happy New Year!

Fine looking work, Mark!

Hooray for finished projects! That's cute, Paul. Since you didn't get to premiere it this Christmas you can use it next year!

I gave it a try. The core problem is that the +x side isn't exactly mirror to the -x side. There are two solutions... -In model mode select all the CPs on the +x side, on the "untitled" Group that creates in the PWS, choose "Snap to mirrored points" or -in the MirrorBone Wizard, raise the "use tolerance for cp position" parameter. I had to raise it to 0.5 to get it to catch everything. MirrorBone creates a Group, "CPs without mirror" of points it couldn't find mates for to let you know if there are problem CPs.