Render foreground figure to different image?
SaraT
Posts: 58
Hello,
I wonder if there is some way to render the foreground figure(s) to one image, and the environment to another? This to make it easier to make smaller composition adjustments afterwards in GIMP/Photoshop without having to do a new render.
In particular, I can't hide the environment and render the figure by itself, because then I don't get the shaodws/lights from the environment on my figure. I'd like everything to cast shadows on everything else, but that those shadows go to different images. In the case of ony one foreground figure, something like:
Image 1: Figure and the shadows cast by the figure on the environment. Figure has light and shadows from the environment on it. The rest is transparent.
Image 2: Background environment and the shadows/lights from it. Shadows cast from figures not included.
Image 2 I could easily do by just removing the figure and rendering the environment, but how to do image 1? Is it even possible? (I'm using Iray.)
Thanks!
Sara
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
I used to do it that way, was always a lot of frustration. These days I just try to get it right in the studio camera. The amount of tweaking I had to do in post when I chopped up images like that usually ended up taking more time than it takes to get it right in the studio camera. It's faster to do a 500 by 500 pre-render to check for stuff that looks wrong, fix it, repeat until it's ready, then do the final large render. At least that's my experience so far.
You can do that using canvases :)
Thanks for the advice! Could you share some more details on how to use them for this case? The tutorials I've found for canvases were more about combining the strength from different lights afterwards. I've tried them for this case, but I only manage to either produce canvases that contain too much (i.e. figure and background, visible once I change exposure), or strange nearly-black canvases that I don't know what to do with.
I agree that pre-renders are great, and I mainly use that, but right now I'm working on a book cover where I'm redesigning the typography, and I need to make some changes to the image because of that (to make the overall composition look better). It took three days to render the scene the first time. In this case, I'm going to have to re-render it regardless, but I'd like to be prepared for future changes, and for future covers. If I'd had the figure on a transparent background, I could have quite easily changed it without having to wait for rendering at all. Or I could render just the figure, at least. (Or even better, a part of the figure, if that's possible with canvases.)
Some of the output files need "conversion", to be used. They output values that are not available in standard "image formats". They may only show as black or white, for various reasons, as the program is unsure what to do with the data, or the data has not been "normalized", for your use.
EG, the depth data uses floating-point values, grey-scale, which is within a set parameter that may not be relative to your items in the scene. Thus, you may have to adjust the min/max cutoffs and translate those "ranges" of floating-point numbers into shade-values, of which, there are only 256 shades in RGB and twice that, still represented as 256 shades, in 512, 1024, 2048, 4096 greyscale pallets. (Unless you have a professional monitor with the extended shade-values, which few people have. Windows and art programs just add "noise", based on the tween-shade, to reduce "banding", when it displays those gradients. Seen if you actually photograph the pixels and compare individual pixel light data, on a typical consumer RGB monitor/display or TV or phone screen. Only a 4/4/4 {zero color gamut compression}, I believe, is honestly capable of true color display, of the actual values. Still limited to the displays ability too.)
NOTE: The highest color gamut I can get on my Titan-V or Titan-Xp is only 4/4/2, with 4K on a double-data HDMI cable. That is borderline 1024 shades of grey, wich is still "noisy". (This is on a 4K samsung, 65", curved screen, which only has one HDMI input per display chanel. You need 2 double-data HDMI cords for 4/4/4 at 4K. But only 1 for 1080p, where I still see 256 bands of banding on 4/4/4 output, for 1024 levels grey.)
I imagine the colors are the same too, which photoshop can sort-of manage, but it still has limitations. 32 bit and 64 bit color is still a potential issue, as there are only some tools that work with 64 bit color still. Most are limited to 16 bit and 32 bit color. (Including layers).
I would say... a second or third render may be unavoidable, for your final desired output.
Exactly what elements, from your render, are you trying to alter?
I'm moving the foreground figure a bit and changing the hair shape a little. I can rerender the center of the image to get it (avoiding background figures), but I'm trying to figure out if there's a better way, especially for the future.
Yeah, image one can be done sorta with canvas, you simply make a node that contains the hero, with their clothes, hair etc. The cast shadow is the issue, there might be a way to get a hsadow pass out of canvases, but I don't know it. You would also still need to do a total seperate render with no hero at all if you need freedom to move the hero around. If you just do a canvas for the backround, there will be a blank spot where the hero was at render time, I hope I explained that right lol.
With canvases you can render the figure without the background, but still have all the lighting/reflections/shadows that come from the environment.
There is a good tutorial for that by Snow Sultan: https://www.artstation.com/artwork/GX30X4
I see. Bummer, but thanks for the explanation!