Looks solid, it might be nice to see also renders with uniform clay material - while you're looking at it you can analyse shading better. You can try to introduce some differences between pieces made of different materials for instance broader bevels on plastic than on metals etc. These are very small nuances but help to make forms and materials more readable. Perhaps some anti-slip patterned plastic could make some parts of grip more interseting. A problem with second render is lit part of the floor in the corner being the brightest spot, try to make it so composition is focused on model having areas of light and shade in mind. I'd suggest looking into three-point lighting setup for start.
Hi Polycount! I thought to share my recent likeness study here too. I Practiced recreating Cassian Andor (portrayed by Diego Luna) with a real-time render in Unreal Engine 5. Artstation link: https://www.artstation.com/artwork/8BG5E6
Thanks for looking. Even though I am finished with this project I am still happy to hear critique.
"Did a lot of research before giving up web dev. Really wanted to make a
living makin flashy sites but with AI i don't see a future there. "
I might suggest another pivot toward CG programming or back-end IT stream, studios are still on the look out for tech artists/game engine software engineers...etc despite this current ai hype, I mean no one in their right mind would allow a bot let loose amidst their inhouse codebase...
Just a personal observation from someone who was in your shoes during the dotcom boom mid 90s to early 2000s era, when I gave web dev a crack though hadn't the necessary attributes to exel at it but occasionally do still keep my hand in on minor stuff.
In my experience, Ai tools make sense to use as a shortcut for small bits of menial work. As long as the task is well-defined with plenty of decent-quality source material for the model to pull from. And only if I am able to filter the output using my own specific subject-matter expertise.
It's great at handling the filler, not so great at getting the specifics correct.
I recently was handed a writeup, created with AI, about my artwork. It took care of around 80% of the formatting and content, but failed miserably at the crucial meat of the post, the most important information we want to share with the audience. Without human editing on the result, the output would be totally incorrect.
Does it save us time? Sure, if we're creating content that's been made a thousand times before, and just needs to be tweaked to match the current subject.
Curious how this is going to play out with actual 3d content-creation work, and not just writing press releases. I suspect it'll end up in the same general area, great to speed up some of the drudgery, but always needing an adult at the wheel to steer us away from the cliff's edge.
edit... I still have major qualms about using AI models at all, since they're all sourced with tons of copyrighted content, without any author permission. It's still all so fucking wrong.