Russia
Joined: Aug 13, 2022
Post Count: 613
Status:
Offline
artificial intelligence to create a model
Dear Keet, dear Captaincook, I have very bad news for you. I'm not smiling anymore. Dear Dorin, in a related topic, said that 10 years ago he had already seen a program for creating 3D photographs of a person's face. And now these 10 years have passed. First, please look at these SH3D renderers.
There are tears on my face. Because now I feel helpless. What seemed like a distant future to me has been here for a long time. And trouble is knocking on the door right now. I do not know how to feel like an artist now and look forward confidently.. Okay, I need to calm down and tell you everything. First of all, you need to understand that now you don't need to look for 3D models of all these exclusive sofas, armchairs, tables and cabinets. All you need is a photo from any angle on a white or other flat background. And that's it, nothing else. Then just watch the AI work. This invisible employee has already learned how to do everything better than a human. Horror.. Here's a picture of the chest of drawers.
Now, if you are standing, then sit down and hold the chair tightly with your hands. Here is a 3D model of this chest of drawers that appeared through (are you sitting?) 15 seconds. 15!!!! seconds. I still don't believe it..
the handles even stick out, just like in a real chest of drawers. It's easy to calculate that with finger dexterity, you can easily make two models per minute. Here, for example, is a photo of a bronze candelabra.
and here's the 3D model in 15 seconds.
A model can be made from anything. You can make a cat, a human. And the most important and scary thing is that AI makes the back invisible shape with the right texture. Here's a picture of the chair. There was only this picture! I'm not lying.
And then 15 seconds later
I don't know about you, but I'm already scared. Especially for Captaincook, even though I feel like a scoundrel and hit below the belt. But you need to know this!! And understand who can open the door and who can't. There's a knock on your door right now, careful. They're already at your door. Photo of a sailboat.
after 20 seconds. Captaincook, please don't blame me. Just don't open the door. They know what's in the stern..
These are the terrible discoveries I have just made. If you need to, I'll continue. Now the monologue is difficult for me. I am waiting for feedback from the community..
Russia
Joined: Aug 13, 2022
Post Count: 613
Status:
Offline
Re: artificial intelligence to create a model
hi. This is Trellis . Registration is required, then you can work without a quota. The model is saved in GLB format, which must then be converted to OBJ. It is necessary that all three files are inside the ZIP - PNG, MTL and OBJ, not just OBJ.
Netherlands
Joined: Apr 8, 2022
Post Count: 1487
Status:
Offline
Re: artificial intelligence to create a model
Dear Keet, dear Captaincook, I have very bad news for you.
For Sweet Home 3D users it's good news. It adds another option to get exactly the 3D model you are looking for even if you can't find an existing 3D model and can't create it yourself. Professional creators will not be happy with this, now it's possible for every amateur to create great 3D models.
The only caveat that remains is the single texture for the complete model. For simple models with mainly colors that can easily be fixed using Blender. For complex models with multiple complex textures it will be very difficult if not impossible to split the materials.
Maybe it's an option to "fix" photos with multiple complex textures before conversion to a 3D model. Edit the photo and replace the different textures with solid colors. After conversion this model is much easier to fix in Blender to assign separate materials. Then you can set appropriate textures in Sweet Home 3D.
How does a character (person) convert to a 3D model? A great way to get your favorite celebrity as a 3D model If the trick with separating materials works it becomes easy to change the color/textures of the clothing, maybe even the color of the nail polish.
---------------------------------------- Dodecagon.nl 1300+ 3D models, manuals, and projects
Russia
Joined: Aug 13, 2022
Post Count: 613
Status:
Offline
Re: artificial intelligence to create a model
so far, the model detail is very average. I am considering this method only for furniture. For example, the person in the photo in excellent detail on the AI model will be like plasticine. But you can put it somewhere in the background)). But you need to be ready - another year and that's it. There will be a model in 20 seconds with all the eyelashes. When photographing from the back)))
Netherlands
Joined: Apr 8, 2022
Post Count: 1487
Status:
Offline
Re: artificial intelligence to create a model
For example, the person in the photo in excellent detail on the AI model will be like plasticine.
I wonder if that is deliberate, to avoid lawsuits from celebrities. I'm sure it recognizes that it is a person and then uses different settings so it doesn't create exact matching faces. Maybe changing the settings will get a better result, for example increasing the sample steps. Or changing from stochastic to multidiffusion (I have no idea what those mean )
---------------------------------------- Dodecagon.nl 1300+ 3D models, manuals, and projects
Russia
Joined: Aug 13, 2022
Post Count: 613
Status:
Offline
Re: artificial intelligence to create a model
I see such opportunities using people in the interior. For example, there is a photo of a girl on a Viennese chair. With the help of a smart camera, you can select a chair and search for a picture to find exactly this chair with its photo. I did it. And then you need to create a separate model of a girl on a chair and you can just put the same chairs next to it. A great idea for the interior.
Russia
Joined: Aug 13, 2022
Post Count: 613
Status:
Offline
Re: artificial intelligence to create a model
I made a mix of an AI model and a high-quality photo on the box. Now it's very easy to get a shape using artificial intelligence, but put it in second place. This secondary model should cast shadows. The first role is played by a high-quality box with a photo, which needs to be interestingly illuminated with hemispheres. This backlight can make a spectacular imitation of the volume in a photo. I made a model of the sofa using AI photography, and it turned out great. And I put two human models on this sofa (they were also made with the help of AI), which of course looked disgusting. But in front of them, I hung a high-quality curtain of photographs on a box in the space. The main task is to combine the bad AI model and this photo from the front as much as possible. Such a tandem can look more natural on a render, with shadows that are difficult to obtain on a person using a group of primitives.
seconds. I still don't believe it..
