🤩 30% more speed and increased quality in high-resolution image upscaling I've improved clarity upscaling model by reducing the creativity by 20% in every internal upscaling step. It makes the amount of hallucination much more controllable, thus increasing the quality while also being 30% faster ⏩Now live: App: Code:... Replicate: 🏞️ It's great how the blurry background stays blurry, but the cat becomes super sharp This example was made with a creativity setting of 0.25, and you can go lower if you want less hallucinationshow more

Please DM me if you want access to the beta ClarityAI API

can i use this on sd forge or only a1111?

Yes, check my GitHub repo for params

LOL thats a completely different cat .

In most AI images, a hallucination is welcomed because it can fix artifacts and deformations. If you don't want to change the image, you can just go very low with creativity.

Wow again amazing, what is the benefit of using your API vs replicate?

Replicate is great if you want something very technical where you can tinker around and find the best parameters by yourself for your use case. My API is very easy to use because I do all the optimisations in the backend, like lower creativity for higher resolutions and multiple presets, and I constantly improving it and I will add premium API features in the future that might not be on the Replicate model.

Would you say clarity is better for increasing definition or fixing malformed generations (i.e. malformed limbs, fine details in things like food that aren’t generated properly)?

You can do both: low creativity -> increase definiton high creativity -> fix malformed generations

And when will you release the improvement open source?

The improvement is already live: