KNN-Diffusion: Image Generation via Large-Scale Retrieval · HackerPapers discussions

KNN-Diffusion: Image Generation via Large-Scale Retrieval — mentioned by gwern

Tue, 07 Feb 2023 01:24:37 GMT

> The AI does NOT build or use reference boards for specific prompts. Interestingly, even if it did build reference boards, that doesn't mean that the generated image would 'copy' or 'look like' the references all that much. We know this because there are diffusion generative models which do exactly that: they do retrievals first, and include the 'examples' along with the prompt before generating. Here is a visualization: https://arxiv.org/pdf/2204.02849.pdf#page=19 (the generated sample on the left is clearly semantically correct and high-quality but also very different from all the…

Paper and discussion history. A mention is not an endorsement.

KNN-Diffusion: Image Generation via Large-Scale Retrieval — HN submission

Mon, 24 Oct 2022 09:51:49 GMT

KNN-Diffusion: Image Generation via Large-Scale Retrieval

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

KNN-Diffusion: Image Generation via Large-Scale Retrieval — mentioned by gwern

Sun, 01 May 2022 14:18:48 GMT

> It doesn’t go out and google for reference material based on the prompt. And note that even if it did (which is definitely architecturally possible given progress in retrieval and demos like WebGPT), it still wouldn't necessarily be 'copying', in the goalpost-moving excuse. Facebook has a neat diffusion model which uses retrieval on a dataset of images for generation: https://arxiv.org/abs/2204.02849#facebook You can see the exemplars it uses: they both greatly improve the quality of generated samples, but also don't look that much like the generated sample.

Paper and discussion history. A mention is not an endorsement.