3D model to AI render in less than 30 seconds.

3D model to AI render in less than 30 seconds.

Today we will learn how to generate great looking images that look like renders made out of your 3D scene. So we have to assume, we have some 3d models, a 3d scene maybe. We do not have textures or light setup yet, just the outlines mostly. And we would love to have a quick glance on the future result, we can simply generate those images in on of our tools. For example in Interior designer, which might work very well for this task:

All the settings you might need are on the screen above. You can pick just edges or image and edges, if your image is a colored drawing it might help, but most of the time you will want just edges in order to not give the AI a wrong idea about the real colors and setup of the model. In the prompt field below the image I have added one of the 28 existing presets for the real estate industry, as it fits well with what I tried to achieve here. Below you can download exactly same 3D model as I used for this example:


How do I turn my 3D image into AI Render with deep-image.ai?

First of all, you need a screenshot of your 3D drawing, 3D scene, or 3D model. You can also make a depth map to get even better results, but it's not a requirement, rather an extra thing to do in case you don't get the default results on a good level.

Then you go to our  AI Generator PRO, or use the Real Eastate Designer, which is easier to use, offers less options but is still a powerful option to use.

Next step is to upload your image. Once this is done, you need to choose the Edges, or Image and Edges in the options (we explained the difference above). The strength slider I suggest leaving on default value (50%), and only changing it after the first generation needs corrections. You will most likely do very well with this  setup, but if not, then feel free to try out various settings of the slider etc.

Now let's add a Prompt. Here you can write something yourself, like a description of what you want to see. For example you can describe a fantastic world that this scene/model should be set in. For the image below I used a prompt that goes:

„The palace of Orcs from Lord of the Rings, it's all dirty, smokey, and sticky. ”
A quick tip: Try using existing presets, and then editing their content to better fit your needs. Often that's the faster way than writing everything from scratch.

After Generating an image, we can download it to our device with a press of a button. And it's done. The average waiting time is between 1-3 seconds, unless it's really busy.

Because it's our 3d model, or it's a free 3d model, we can play around. I took a virtual walk inside the building, found interesting angle and decided to reimagine it:

That's my input image from the model

I used one of the existing presets to reimagine this as an industrial loft:

Of course you can Reroll the image if you didn't like something and get another one.

As you can see, even though my input image was quite dark, it was actually difficult to make sense out of it, we still got a stunning result with correct lighting reflections, realistic textures and correct perspective. It's quite impressing, especially if we assume that AI does not fully understands what it draws. It still does a good job. If you wish to have a look at this exact example, please use this link to the result photo. And then click "Back to Edit". You can exchange an image for another one, and just click "Generate" leaving the rest of the settings untouched.

If you are looking to make some yourself, open AI Generator PRO or Real Estate Designer now and try to follow my instructions step by step. You can experiment with various visual styles, you can name a specific game to impact your images with it's aesthetics.

How rendering with AI can be more useful than a normal render?

Simply put, you can modify your prompt and immediately get 20 drastically different styles, colors, textures etc. When working with a normal 3D editing tool, you would require to set up whole scene lighting, retexture every single object, and render it all again to get a similar result. So it's an excellent shortcut, especially on the conceptual phase of the project, where you want to go through many iterations very fast to check possibly big number of versions.

You save plenty of time, and give your team content that they can use to make better decisions. You can also get inspired, as sometimes AI can draw really cool things that might be useful.