Iray GPU rendering, scene starting tips

todorivanov09

I started to do the tutorials last night and I have a few questions.

1. How to enable Iray to use my GPU, not my CPU?
Right now I have enabled both graphics (in the BIOS) - dGPU and iGPU (using it for my IPS 1440p monitor, which also runs the program)
My GPU is more than enough for the tutorials for sure - Asus Strix 1070ti (factory OC)
The problem with the CPU rendering is - my PC becomes unusable for 30 seconds. Clicked on every window where it had "Render Settings" but couldn't find it.

2. What is the correct way to start a scene?

Background -> Model -> Clothes
Model -> Adjusting it -> Clothes -> Background
Model -> Clothes -> Adjusting it -> Background

Currently, I created my model, adjust every single thing, that I wanted, but when adding clothes, that should fit this model, they look like crap, the collision is broken. Tried right click and "Fit into the model" (or something like this) but nothing happens. The same goes for the hair and mustaches/beards, my character is looking like after four sessions of chemotherapy. I am not sure if my adjustments, to the body, broke them or something else.

Kitsumo

In Render Settings, click the Advanced tab. From there you can enable/disable CPU rendering.

There are only 2 ways to do a scene: 1.) Your way  2.) Everyone else's way   

In other words, just do what works for you. I like to model the characters first because it's easier to move them with nothing else loaded in the scene. After I have the characters (and maybe furniture if they're sitting), then I add clothes, then background. Then I make any cosmetic adjustments, morphs, adjust textures, etc.

JonnyRay

My flow for scene creation goes like this..

1. Load the base character into an empty scene
2. Apply morphs and shapes and skin shader adjustments if necessary while they are still in the default pose
3. Keeping the character selected, load the hair I want on them
4. Adjust sytles and colors as necessary, but I do not pose it yet.
5. I will do some quick test renders with a standard HDRI that I use for character setup from Colm's Pro-Studio HDR Lighting System
6. At this point, if I know I will re-use this several times (I tend to re-use certain characters a lot), I will save the figure and hair as a "Scene Subset" so I can easily re-load them again.
7. With the character selected, load the clothes I want them to wear
8. Adjust the fabrics and surface settings
9. If I know what pose I want for them, I will set the pose and make some adjustments to clothing and hair as necessary for as much of a natural look as I can get
10. During this stage, I will often do more test renders using the same HDRI lighting since there are some things that only show up during a render
11. When I'm roughly happy with the character and pose, I save the scene (usually calling it something like "Conall-setup")
12. AND I save a scene subset with only the character, hair, and clothing. The reason for this is that Scene Subsets don't include the render settings; so I can merge them more easily into my final scene without messing up any HDRIs I'm using for my background and final lighting.
13. I will repeat the above for each character in the scene
14. Then, in a new scene I load my setting, populate it with furniture and other props and I get the sky dome / hdri lighting set in roughly the way it should look
15. A couple quick test renders will let me know if I have major issues like a building casting an unwanted shadow over the spot where I want my characters to be.
16. Then I merge in the first character and position them roughly into scene
17. Assuming I have multiple characters interacting, I'll import each one, get them rough set, then import the next
18. Now I'll setup my camera to frame the image. I typically setup my aux viewport to use that camera so I can always check how things look from the camera's perspective as I start to work on the details.
    1. Framing the camera also allows me to only work on details that truly matter. If a character's toes poke out of their shoes, but the shoes aren't going to be anywhere in the final image, I can leave them alone.
    2. Similarly, if one character's hand is technically slightly inside the back of another character, but the image still looks okay from the camera's point of view, I won't bother fixing it.
19. I make all my final posing adjustments at this point. Especially getting their face and eyes to look the proper direction.
20. I may move around props and such as well if they work better from their default positions.
21. I also turn off anything that isn't going to affect my final scene. For instance, if there's a tree behind the camera or a bush off to the side that won't be seen and doesn't cast shadows (that I want) on my image, then I hide it. No need to send extra data to the rendering engine if it isn't providing value.
22. Now I start my final test renders for lighting adjustments
23. Once I'm happy with the lighting, I'll render my final image. Usually to Iray canvases so I can do some postwork if I need to.

I admit that my process can take a while. Especially with all the test rendering I do. But I feel like working in pieces like this allows me to work on "scenes" that are smaller and faster to work with and render than if I tried to do everything I just described within the final setting.

LenioTG

• I create the characters separately and I save them as scene subsets.
• I set the environment before everything else, lighting it up the way I want it, and setting the render settings. I save it as a scene, in case I wanted to use it again in a similar render. I do stories, so it often happens that they go to a characters house or to their workplace, for example. For the lights I mostly turn down the environment intensity and use the Ghost Light Kits 1 and Probe. The 1 for strong realistic lights from windows, and a weak one from the ceiling. The the probe give a more poetic effect and light up the characters faces too, speeding up the render.
• I decide where the characters are going to be, and I select everything (ghost lights too) and I move that spot to the center of the axis, to avoid black eyes and hailine problems.
• I merge the characters into the scene. I start to pose them (final touches can be made only when they're dressed). The pose order is full body->space position->expression->look at me product->final adjustments (after the clothing part).
• Test render, if needed I add some ghost lights to light up the character, or some light props from LI Incandescent set. But I hate artificial lighting, so I try to keep it simple. A good environment lighting should be enough 90% of the time.

In my YouTube tutorial channel I've actually made a video with my complete workflow. It has improved a lot since then, but I think it's still a viable video.