Anonymous at Sun, 26 Nov 2023 16:00:32 UTC No. 965405

You don't and you can't because you don't know shit about 3D and art in general.

Anonymous at Sun, 26 Nov 2023 23:34:05 UTC No. 965417

>I import the SDXL image into an orthographic view in maya and then place a flat plane over it.
>I move the verts and make new edges if need be to match the subject.
>I then push the verts back and forward only to add depth

incredible. can I see the topology so far anon, I promise I won't bully.

Anonymous at Mon, 27 Nov 2023 01:07:53 UTC No. 965424

>>965417
i modeled this about 8 months ago with that workflow. These are all renders though. The stable diffusion reference is not shown.

Anonymous at Mon, 27 Nov 2023 12:32:43 UTC No. 965452

>>965402
That type of thing is only possible in blender sorry anon

Anonymous at Mon, 27 Nov 2023 21:10:16 UTC No. 965479

>>965402
can't these things also make the 3d model now?

Anonymous at Mon, 27 Nov 2023 23:30:41 UTC No. 965487

How do you understand the word orthographic but can't figure out how to solve your problem?

Anonymous at Tue, 28 Nov 2023 01:22:53 UTC No. 965496

>>965487
what did you mean by this, senpai?

Anonymous at Tue, 28 Nov 2023 01:54:38 UTC No. 965497

>>965496
it's a very simple spatial reasoning problem you're having. One that I don't think you should be having if you understand the term orthographic.

Anonymous at Tue, 28 Nov 2023 02:14:21 UTC No. 965501

>>965497
I made this >>965424 with that workflow, 8 months ago. Now I want to apply it to more recognizable stuff. What do I have to do, senpai?

Anonymous at Tue, 28 Nov 2023 02:44:58 UTC No. 965503

>>965501
You need to use a different workflow.

For a start, orthographic images don't show depth! So you're moving a vert around to look roughly correct, from a particular view, but there's a million ways for it to look correct from that view and one way for it to actually be correct. And it's literally just luck.
But beyond that, there's the reason why basically AI images are terrible for artists to use as reference. I'm totally against AI, I don't use it in my work at all, but at most I could say you could use it for inspiration. But inspiration isn't reference. You use references to find out what something actually is in reality, or for learning how another artist solved a problem.
You can't use AI for that, it gets things wrong. And you don't know what it got wrong because if you could tell you wouldn't need the reference to begin with. References are for learning and you want to learn from high quality sources.
With you it's even worse because you're using a workflow that demands you see an object from multiple perspectives. Tracing an image in 3D like that is a valid workflow if you have the object from multiple perspectives or are willing to be creative for the other perspectives. AI can't give you multiple perspectives like that consistently.
So if you want to model more recognisable stuff, you're just going to have to use high quality references for the recognisable stuff. Get pureref, go on pintrest and google images and shit, find lots and lots of images of what you're trying to make, and just make it.
Use the AI as inspo if you want, but don't use this workflow it's no good.
Your way is maybe quicker than some other things, but that's a fool's economy. Put more work in and you'll reap the rewards in the long-term by being not shit at this. It's more fun too.

But bottom line, your workflow is no good because you're trying to work with information you don't have, and you want the computer to do something with information you haven't given it.

Anonymous at Tue, 28 Nov 2023 02:55:35 UTC No. 965504

And btw the reason this is so funny and such a simple problem is because you're working in 3D. Just orbit around the object as you work to make it look good, and use your artistic judgement.
if you just actually make the object in 3D space like a normal person, the problem goes away.

People basically only use similar workflows to this for like static FX shots where you know 100% the perspective isn't going to shift. It's a quick and dirty way to work, but it's good enough for certain scenarios.

Anonymous at Tue, 28 Nov 2023 03:45:33 UTC No. 965507

>>965503
>For a start, orthographic images don't show depth!
my image isnt orthographic, its a random perspective image from SD

i use an orthographic camera to move the points of the plane in XY and then use the perspective viewport to move in Z

🗑️ Anonymous at Tue, 28 Nov 2023 15:02:17 UTC No. 965545

>>965507
bump. What is wrong with this method?