SU/ LO/ AI tutorials you didn't ask for

[bmaxim]

Admit it, how many times have you thought about making a video using LayOut? Exactly, me too, every day

Given everything happening in the world, I wanted to keep my mind engaged with something else. What could I do besides making videos in Layout?

I got a hype video from YT (don’t know what he’s singing about, I don’t know English well) + I have a collection of 2d People models - let’s get to work

The main character

• Set the video to a lower frame rate, ( in my case, 5 fps), and export this video as .jpg images.
• Create as many layout pages as there are frames (in my case, about 300) and place each frame in order on each page.
• Manually draw the main character’s profile separately for each page.
• Delete the imported .jpg images, set a green background for the chroma key.
• Export the images as .png files and import them into After Effects (or a similar program) as a .png sequence.

Background
I had a few ideas, but I ultimately decided to export SU animations and experiment with them in After Effects. Here’s my plan:

• Find a stadium on 3DW and arrange my collection of 2D people models, similar to what’s shown in the video.
  

  

• Plan to play with styles, lines, materials to create a more minimalistic look.
  

  

• Make a general scenario of the video in my head and based on it export the video animations from SU to AE

• Make some AE adjustments

The final AE panel looks like this:

So, what is the result? (low-quality video due to forum size limitations) :

Congratulations, if you had a weird week, I just made it even weirder for you

[ateliernab]

you just reinvented Rotoscoping

it’s like people doing art in excel. it’s not intended for that use, but if it works… well it works.

[mihai.s]

Ingenios, Max!

[bmaxim]

I noticed this topic became increasingly popular , so I continued developing it. This topic will likely focus more on AI.
What if I created a song describing the making of this video? How good is AI at creating a song? Well, you can decide for yourself:

• I copied the description of the post above and pasted the description into ChatGPT’s Suno AI v4 for music and song lyrics generation.

• I copied the final lyrics and pasted into the AI Song Generator https://ai-song.ai/

• I added the final clip (with the new audio) to https://echowave.io/ to add subtitles

The final result!

[StudioSKP]

Absolutely brilliant. Nicely done.

Cassidy
PM Graphics SKP

[bmaxim]

How do I create professional renders quickly to achieve a “wow” effect when I’m in a hurry?

I’ve been testing different methods to get high-quality renders at the very beginning of a project If you’re using SketchUp + SU Diffusion + ChatGPT, here’s a workflow that’s been working for me:

1. Export your SketchUp view
Take a clean screenshot or export of your scene then upload it to ChatGPT

2. Use ChatGPT to generate a professional prompt for SU Diffusion
Instead of manually writing your own prompt, just ask ChatGPT to do it for you — it usually gives you a very clean, optimized prompt tailored to your scene’s style and goal.

3. Paste the prompt into SU Diffusion
Drop that prompt into SU Diffusion and generate. Pick the one that best captures your vision.

4. Upload that render back into ChatGPT
You can describe what you’d like to improve — Add people, blurred cars, corrected lighting, or more atmosphere. ChatGPT will help adjust the prompt to refine the render further.

5. I often rewrite the prompt if the rendering is too distorted, or I can add other tweaks

6. If you have time, import the final renders from both ChatGPT and SU Diffusion, then blend them in Photoshop for even more control and polish.

---

• PS: You can add the rendered image from Chat GPT in SU Diffusion to ‘‘Add a reference image’’ - it makes it even more realistic. Try to Experiment .. :)*

• PS2: Impressed by SU Diffusion (but more so by Chat GPT). I still haven’t figured out how to properly use artifacts and regenerate parts (I assume these options will be dedicated more to SU Ipad)

[gsharp]

We need a AI category in the forum .. just a thought !

[bmaxim]

Continuation of the post above: A comparison between ChatGPT, Google Gemini, Grok and Copilot.
Same render – same prompt.

The original SketchUp Diffusion Render

The prompt:

Respect model geometry and preserve material/ colors of the building Add realistic trees around this render, spread some pupils and a few adults walking on the concrete pavement in front of the school. And a car on the asphalt in the foreground of the image. Make the car and the walking people blurred.

Chat GPT

Gemini

Grok

Copilot

Copilot – stop it, you’re scaring me!

[bmaxim]

It took me some time to find an AI rendering engine for elevations. I tried many, including SU Diffusion. None of them handle it well—they don’t understand elevations and end up generating perspectives instead of parallel projections. Only ReRenderai. com (below is an SU/ReRenderAi example) seemed the closest to reality.

Has anyone tried rendering realistic elevations/ paralel projections with any AI renderers, since ReRenderAI is quite expensive?