FINALLY tasteskill supports dashboards and app screens too! here's an example

here is the prompt + reference images github.com/Leonxlnx/procedur… Recreate the environment in these 10 reference images in 3D, using only code. No assets, no textures. It's a big environment: a lake with mountains and forest around it. The task has two parts: 1. Build the 3D environment. 2. Make a cinematic one minute video that showcases the world, and render it. The environment: - Make it look 99.9% like the reference images. Don't stop early because you think it's good enough. Iterate until you reach 99.9% similarity. It might take days, I know. There's a lot to do. - Colors: 100% clean, the best colors. Compare your colors with the colors in the images. Most of the time the colors come out wrong and a bit too grey, and coloring is one of the most important things. Light and shadows have to match too. - Water: get the water right. The colors, the physics, everything should look perfect. - Trees: they can't look fake. Give them much better textures and make the forest denser. - Mountains: better textures and a better vibe. - Rocks and terrain: good rock textures and a lot more detail in the small stones and the terrain. - Distance: render everything. Don't render the trees only half of the distance. Trees far away need detail too, also in the half-wide shots. - In the end it should look like a real place to the human eye. The video: - Cinematic, like a drone-filmed movie. That's only a comparison: no drone UI, no drone noises, just a clean video. - Start with a really good scene and show the best stuff first. Then slowly show more and more: for example, you're behind a rock and can only see a bit, then the camera flies up and you see a really nice view. - Every six or seven seconds, switch to a different place. Sometimes fast, sometimes slow. - Sound: calm nature sounds and epic background music. The music has to be really good. - Output: high resolution and ultra-wide, so you can see the trees on the other side as well. Clean and not buggy. Keep full quality, even if it takes longer. How to work: - I don't have a strong GPU, so keep that in mind. You can use my GPU, on CPU it would take a lot of time. - Work in rounds and send me screenshots after each round, without pausing. If it still needs improvement to reach 99%, start the next round right away. - Don't start the final video before everything is really good. Send me the screenshots first, wait for my approval, and only then render. - Don't stop until it's really done and really good. If you have questions, ask me now. Otherwise, build as long as you can and make it awesome. Go and cook.

github.com/Leonxlnx/cinetic this is the cinetic skill I'm using to create my videos btw!

millions is ofc an exaggerated number, I just think that the idea would be cool. showing the anti ai people that AI is not like chatgpt in 2023 anymore