Octane render pass (OBJECT ID) in DOF is slightly thinner...

Forums: Octane render pass (OBJECT ID) in DOF is slightly thinner...
Generic forum to discuss Octane Render, post ideas and suggest improvements.
Forum rules
Please add your OS and Hardware Configuration in your signature, it makes it easier for us to help you analyze problems. Example: Win 7 64 | Geforce GTX680 | i7 3770 | 16GB

Octane render pass (OBJECT ID) in DOF is slightly thinner...

Postby Ofcourse » Tue Nov 14, 2023 4:20 pm

Ofcourse Tue Nov 14, 2023 4:20 pm
Hello there,

I have really big problem and now comes the time to post first question about it... I have Beauty pass which is in small DOF... part of an image is blured and I need to mask some object in this scene and for this I use Render Pass Object ID (Object Tag is assigned on specific object)

But the problem is that the object ID pass has those blured edges slightly less blured or how to say that... Please see the gif attached here. I cannot to do proper compositing because it makes the composite image with edges not being masked out as it should be.
https://drive.google.com/file/d/1NfgjzpF_DCtztMZFEc4Nu07RKcS3sBnh/view?usp=sharing
Ofcourse
Licensed Customer
Licensed Customer
 
Posts: 19
Joined: Mon May 06, 2019 6:37 pm

Re: Octane render pass (OBJECT ID) in DOF is slightly thinner...

Postby jan8 » Wed Nov 15, 2023 8:20 am

jan8 Wed Nov 15, 2023 8:20 am
couple of semi helpful (compositors) thoughts -without knowing anything about your setup and most importantly your viewing environment and goals with it:

- you *might* be applying a LUT/tonemapper to your beauty but not to your utility pass so any gradation will appear to fall off differently (most viewing env would/should do so)

- MOST importantly, there is no 'correct' way to apply effects or cut out an element that has already diffusion (DOF in your case) WITHOUT performing any kind of color unmatting, since any matte/mask will have some kind of filtering applied (even AA would count as such) - this can be really problematic with high contrast composites (or photos).. just think a dark out of focus element in the foreground and something really bright behind it: ANY soft matte would also include pixels of the BG which will always print through - depending on your image, this might or might not be a problem. you might want to start tweaking your matte but that of course comes with a heap of other problems (any of these operations work in screen space).
long story short, what you are trying to achieve is done in production a lot and it kinda sorta maybe works. if you render with DOF baked in however all these problems become much more notable (heavily motion blurred images do suffer from the same)
jan8
Licensed Customer
Licensed Customer
 
Posts: 35
Joined: Sat Mar 21, 2020 7:26 am

Re: Octane render pass (OBJECT ID) in DOF is slightly thinner...

Postby Ofcourse » Wed Nov 15, 2023 11:06 am

Ofcourse Wed Nov 15, 2023 11:06 am
Thank you for your message here, I do understand that there is not a right way to color correct already baked in DOF on Beauty Pass from quite few reasons... I have to note that I do not use any LUT applied on my Beauty Pass and all in my preview cut out is just basic settings in use. So when I noticed the differences between Beauty Pass DOF and Object ID pass (with DOF) isn't match exactly the same... The Object ID is blured slightly differently as you can see in the gif here... otherwise it makes the color correction hard job to do :(
Ofcourse
Licensed Customer
Licensed Customer
 
Posts: 19
Joined: Mon May 06, 2019 6:37 pm

Re: Octane render pass (OBJECT ID) in DOF is slightly thinner...

Postby SSmolak » Wed Nov 15, 2023 1:00 pm

SSmolak Wed Nov 15, 2023 1:00 pm
Generally it looks like any of "filter" that move pixels from original positions like DOF, aberration cause problems with masking.
Architectural Visualizations http://www.archviz-4d.studio
User avatar
SSmolak
Licensed Customer
Licensed Customer
 
Posts: 1095
Joined: Sat Feb 07, 2015 5:41 pm
Location: Poland

Re: Octane render pass (OBJECT ID) in DOF is slightly thinner...

Postby jan8 » Wed Nov 15, 2023 7:53 pm

jan8 Wed Nov 15, 2023 7:53 pm
Ofcourse wrote:I do not use any LUT applied on my Beauty Pass and all in my preview cut out is just basic settings in use. :(


the beauty appears to look tone mapped though, as in any frame buffer - whether standalone or hosted by a DCC - does apply at least a standard sRGB tone mapper.

either way the problem will persist as even anti alias can be considered filtering.
as a debug routine you could render without any AA (and the DOF), switch off the tone mapper and then compare again.
don't get me wrong, I'm not saying your findings are wrong, All I'm saying is there there is a fundamental conceptual flaw in the notion that soft edges can be isolated when the BG is already shining through/baked in. So even when taking a photo and drawing a rotoshape around an object, you end up with the same problem and compositors having to go through hell to get rid of the portions of the BG that is partially visible through the edge.
Vray attempted to mitigate this with coverage map AOVs which would aid for some kind of color unmatting (by that I mean the removal of undesired pixel (values) that shine through, but I have never seen this implemented fully successful in a comp app
sorry I'm not helpful here and much rather harping on a long standing issue.
jan8
Licensed Customer
Licensed Customer
 
Posts: 35
Joined: Sat Mar 21, 2020 7:26 am

Return to General Discussion


Who is online

Users browsing this forum: No registered users and 22 guests

Sat Apr 27, 2024 7:08 am [ UTC ]