View unanswered posts
View active topics
It is currently Fri Sep 25, 2020 1:03 pm


*  Forum rules
Dear users,
we are sorry to inform you, that we had to end development of our FurryBall after almost ten years!
We will give FurryBall for free to all users, but without any Maintenace and Support.
Because we have to keep our servers working, we have set the symbolic price 29,- EUR per year.
Hope you will understand us and keep using FurryBall if it fits you.
Thanks for all those nice years with GPU rendering.
Your FurryBall team

Author Message
ymangolds
Post  Post subject: Request: Deep image output  |  Posted: Sat Jan 30, 2016 6:07 pm
User avatar

Joined: Fri Aug 17, 2012 8:23 am
Posts: 89

Offline
I've been researching/learning about deep compositing since my current project needs massive amounts of volumetrics/voxels and am really impressed with all the things that it allows (can understand why it's becoming an industry standard for the big production houses). Although basic deep data can be created in post using depth and/or world position passes, it's not as accurate and if the scene has transparency there's not too much an improvement over old-school depth pass compositing (since the multi-samples per pixel in true deep data passes is what makes it awesome).

On a somewhat related note, adding an option to output a single exr aov file per frame with all the passes as channels would be nice and make post production tidier.

While I'm requesting a new render pass... :oops: I may as well also request a motion vector pass (camera relative) to make scenes that require complex post production and involve lots of stock elements easier by doing motion-blur in post (even though post-production motion-blur could never be as fantastic and quick as FB's for CG elements ;) on some scene's it'd be nice).


Top
ymangolds
Post  Post subject: Re: Request: Deep image output  |  Posted: Mon Feb 01, 2016 1:33 am
User avatar

Joined: Fri Aug 17, 2012 8:23 am
Posts: 89

Offline
A real (distance from camera) depth pass would also be awesome (similar to how world position pass uses real values). Currently it looks like the values in the depth pass are a percentage from 0.0 to 1.0 based on the relationship of the pixel from the far clipping plane of the camera being rendered (so anything farther then the far clipping plane (which still gets rendered btw) has a value of 1.0). I'm trying to figure out a workaround to convert it into a real depth pass in post (multiply every pixel's R value by the distance of the far clipping plane then shuffle that into the depth.z channel gives values that are like 150% of what they should be), but haven't been able to get anything accurate.


Top
Jan_Smejkal
Post  Post subject: Re: Request: Deep image output  |  Posted: Mon Feb 01, 2016 10:49 am
User avatar

Joined: Fri Aug 13, 2010 10:30 am
Posts: 794

Offline
Hi,
problem is, that depth pass isn't distance from camera! It is distance in view direction and also in non-linear space, so if you look through pixel exactly in the middle of screen, than depth 1.0 means far plane and 0.0 near plane. I think that this distance can't be computed in post process without camera information (far plane, field of view etc.). The simplest way if you know camera position is to use world position pass (length of vector 'world position - camera position' is your distance), however I think, that you don't know your camera position for all frames ;(
We can add this distance path in next release.

Motion vector pass is more difficult to do, so I am afraid, that this pass cannot be added in near future.


Top
ymangolds
Post  Post subject: Re: Request: Deep image output  |  Posted: Mon Feb 01, 2016 8:03 pm
User avatar

Joined: Fri Aug 17, 2012 8:23 am
Posts: 89

Offline
Jan_Smejkal wrote:
Hi,
problem is, that depth pass isn't distance from camera! It is distance in view direction and also in non-linear space, so if you look through pixel exactly in the middle of screen, than depth 1.0 means far plane and 0.0 near plane.

The testing I've been doing is using the center pixel. Specifically, I have several shots that take place inside a Stryker and several of them show the gunner using a new control station for the remote weapon platform. The screen he's using has a view through the targeting camera with a HUD/GUI on top that includes info like "distance to target." So I've been trying to automatically animate that text by using expressions to sample the center pixel of the depth pass.

Jan_Smejkal wrote:
The simplest way if you know camera position is to use world position pass (length of vector 'world position - camera position' is your distance)
That was the next thing I was going to try. I already import the camera into Nuke from the scene that takes place inside the vehicle because I replace the monitors on the control stations (I import the geo for these also) in post and have to do it in Nuke's 3D since the camera inside the vehicles is usually animated and some of the screens are curved monitors (so a simple 4 corner pin won't work). I'll try importing the camera from the exterior scene used for the view on the screen in a similar fashion, and then use d=√((x2−x1)^2+(y2−y1)^2+(z2−z1)^2) from it and the center pixel of the world position pass.

Jan_Smejkal wrote:
however I think, that you don't know your camera position for all frames ;(
For anyone trying something similar: you need to bake the animation on your camera for every frame before export/import.


Top
Display posts from previous:  Sort by  
Print view

Who is online
Users browsing this forum: No registered users and 2 guests
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum
Jump to:   
cron

Delete all board cookies | The team | All times are UTC + 1 hour [ DST ]

Powered by phpBB® Forum Software © phpBB Group
DAJ Glass 2 template created by Dustin Baccetti