I actually have a bunch of videos already bookmarked on how to make a a geode in blender, and how to make a material. But you don’t learn what new technology can do unless you try. But yes, I originally planned to make my own, so I’ll probably do that. But I still learned a lot from this.
3 Likes
You know, that might be a fun use case for Blender’s geometry nodes. I wouldn’t be surprised if someone’s already created one. Not that this would be game optimized either. Parallax occlusion mapping might be another route to look into.
But you’ve already got game dev chops, @dragonforge-dev. I’m sure you know when and where to take shortcuts. My message was for the new people wanting to skip right to the center of the tootsie pop before even trying to take the wrapper off.
2 Likes
Not sure why something like a geode was taken as an example here. Modeling it by hand is pretty much trivial. It’s just a displaced sphere.
Cause he asked for a picture and that was on my mind.
Geometry nodes, or a POM is pretty much trivial as well. The point is that there are many techniques available and new game devs will learn about none of them by having AI do all the work.
1 Like
AI cleanup, photogrammetry cleanup, or learning how to 3d model, it’s all going to take a fair bit of time.
The point is that there are many techniques available and new game devs will learn about none of them by having AI do all the work.
Right. And there is also upskilling, which seems like forgetten concept being because AI is an easy route to results (whether bad or good).
The translation of 2d images to 3d requires some aesthetic/creative judgement. I’ve my share of concept-to-3d work, and what you quickly learn is just because you know how to model something, it doesn’t automatically meant you know to interpret drawn lines in to volumes.
The AI bit in all this bypasses the practice of the skill to translate 2d to 3d, which seems to me at least part of the point of making 3d games? Imo, that’s a loss for folks who want to get stuck doing 3d games, and using 2d concepts as their main engine for ideas.
Even before AI, the messaging from app/plugin developers goes along the lines of ‘we do the “menial” stuff so you don’t have to, so you can focus on the creative… blah blah’. But skill/craft can go out of practice, especially if you have little to begin with; and once it devolves into that, I really wonder what kind of creativity we’ll be seeing.
3 Likes
Only if you care about it. People who eagerly jump on “ai” for this don’t care about aesthetics. It’s the same demographic that’d previously just use what is known as “programmer art”. Just that now it looks much “prettier”… to them.
As much I’d hate to disagree to this, I have to challenge it. Retopo gets easier depending on your tool of choice; ZB, 3DP, Blender, Maya, etc. An AI point cloud is equivalent to one from photogrammetry (not that you used that example).
Retopo requires some knowledge of why need to be topologically-sound. But actually modelling something from scratch requires a lot more skill; it goes beyond the scope of just being proficient in DCC tools/addons or tracing loops.
I also know that everyone has their own priority, but what I’d like to argue is that in learning modelling in 3d teaches you something else: more than technical skill, but perhaps some artistic appreciation of shape or volume, or problems that test you. If you hone your skill, you can carry that with you. If one keeps on relegating opportunities to learn something that is arguably intrinsic to the thing they are making, even though they deliver a product, they’re not any better at doing 3d for the next one.
Only if you care about it. People who eagerly jump on “ai” for this don’t care about aesthetics.
Yes, it is surely about taste. I would like to think that folks who draw concepts as their source material would have the desire to see those concepts realised properly. And perhaps the reason for the eagerness to go into AI is because the road to translate 2d to 3d is hard (I can attest it is hard in a professional environment where each job/client have differing demands and expectations). But that’s why I think it’s worthwhile doing. Unfortunately, fewer and fewer people want to get stuck in to the work they’re doing.
“programmer art”
There are some developers like Jeff Vogel (Spiderweb), afaik, had made a personal effort early on; I think he figured he was never going to be good enough, so he hired artists to do the work instead. He couldn’t improve their own art skills, so he gave someone else the opportunity to. I hope that principle still holds.
I actually agree with a lot of what you’re saying. Learning the craft is valuable, and good artists will always be needed. AI doesn’t replace creativity or experience.
I only mentioned AI because the original question was about converting 2D images into 3D, and that’s an area where open source models are improving incredibly fast. There are already some impressive projects on Hugging Face, and every month the quality gets better.
Interestingly, I used to spend most of my time developing in Godot. These days I work full time with local AI models, integration, and AI governance. That’s actually what I make my living from now, so I naturally keep up with the latest developments in that field.
I still follow the Godot community because I enjoy it, and I thought it was worth pointing out that AI has become another tool in the toolbox. It’s not about replacing artists or avoiding hard work. It’s simply another technology that’s moving very quickly, and for some tasks it’s already surprisingly useful.
Check how your post scores on this “ai” bullshit bingo card:
| B | I | N | G | O |
|---|---|---|---|---|
| Just another tool in the toolbox | Improving at an exponential rate | Namedropping a trendy repo/platform | It is not about replacing humans | That is actually what I do full-time now |
| Every month the quality gets better | Pivot from a traditional tech stack | Passive-aggressive: Interestingly… | Surprisingly useful for X | Staying ahead of the curve |
| Open source will save us | I actually agree with your point, but… | Lip service to human craft or art | The landscape is moving so fast | Unprompted flex about local hardware or models |
| Mentioning AI Ethics or Governance | X profession will always be needed | Overly specific niche use-case | It is about workflow integration | Democratizing the technology |
| Softening a controversial stance | Just keeping an eye on developments | Career pivot justification | Not about avoiding hard work | It is simply the future |
8 Likes
Kit-bashing with Kenny’s Asset Forge tool is another really easy way to get your ideas out into rough/blocked-out models.
I’ve purchased the full program as it’s a one-off payment, but you can use the free version too until you need more pieces.
This means you can get over the initial hurdle of Blender’s learning curve, while still introducing you 3D modeling concepts.
4 Likes
For 3D, learn Blender.
If you use AI you’ll still need Blender to perform extensive cleanup, texture mapping, etc.
If you don’t use AI and create your own models you’ll need Blender.
If you buy models (CGTrader.com, GameDevMarket.net, etc.) you’ll probably (definitely) still need Blender to do minor cleanup, merging animations, re-texturing, etc.
Imphenzia, especially his earlier work, is great for low poly stylized work. Watch his 10 minute modeling challenges. https://www.youtube.com/imphenzia
Grant Abbitt ranges from low poly to fairly complex. Definitely a Blender guru.
https://www.youtube.com/@grabbitt
If you’ve gone through their channels and applied what they teach to learning Blender you should be competent enough to decide where to go from there. Don’t think this is a “weekend” thing either… several months would be reasonable to get good enough that you just know how to get what you want (learned key sequences, what tool to use and when, work flow processes, rigging, etc.) You’ll still be far from an expert but should be good enough to create what you want.
There are many videos of using the 2D image as a template to create your 3D model. For example: https://www.youtube.com/watch?v=bwK7uk-A1y4
You are supposed to begin learning Blender by drawing a donut.
/s
Sarcasm aside, this will always be the correct response 