Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
CLIP + FFT/DWT/RGB = text to image/video
| Date | Stars |
|---|---|
| 2026-07-31 | 789 |
| 2026-08-07 | 790 |
| 2026-08-23 | 789 |
| 2026-08-31 | 789 |
| 2026-10-03 | 788 |
| 2026-10-22 | 787 |
| 2026-11-04 | 787 |
Today
— stars today
This week
— stars this week
This month
-1 stars this month
Momentum
0.0
growth rate 0.00%/day
# Aphantasia <p align='center'><img src='_out/Aphantasia4.jpg' /></p> This is a collection of text-to-image tools, evolved from the [artwork] of the same name. Based on [CLIP] model and [Lucent] library, with FFT/DWT/RGB parameterizers (no-GAN generation). *Updated: Old depth estimation method is replaced with [Depth Anything 2].* Tested on Python 3.7-3.11 with PyTorch from 1.7.1 to 2.3.1. *[Aphantasia] is the inability to visualize mental images, the deprivation of visual dreams. The image in the header is generated by the tool from this word.* **Please be kind to mention this project, if you employ it for your masterpieces** ## Features * generating massive detailed textures, a la deepdream * fullHD/4K resolutions and above * various CLIP models * continuous mode to process phrase lists (e.g. illustrating lyrics) * pan/zoom motion with smooth interpolation * direct RGB pixels optimization (very stable) * 3D look, based on [Depth Anything 2] * complex queries: * text and/or image as main prompts * separate text prompts for style and to subtract (avoid) topics * starting/resuming process from saved parameters or from an image Setup [CLIP] et cetera: ``` pip install -r requirements.txt pip install git+https://github.com/openai/CLIP.git ``` ## Operations [](https://colab.research.google.com/github/eps696/aphantasia/blob/master/Aphantasia.ipynb) * Generate an image from the text prompt (set the size as you wish): ``` python clip_fft.py -t "the text" --size 1280-720 ``` * Reproduce an image: ``` python clip_fft.py -i theimage.jpg --sync 0.4 ``` If `--sync X` argument > 0, [LPIPS] loss is added to keep the composition similar to the original image. You can combine both text and image prompts. For non-English languages use `--translate` (Google translation). * Set more specific query like this: ``` python clip_fft.py -t "topic sentence" -t2 "style description" -t0 "avoid this" --size 1280-720 ``` * Other options: Text inputs understand syntax with weights, like `good prompt :1 | also good prompt :1 | bad prompt :-0.5`. `--model M` selects one of the released CLIP visual models: `ViT-B/32` (default), `ViT-B/16`, `RN50`, `RN50x4`, `RN50x16`, `RN101`. One can also set `--dualmod` to use `ViT-B/32` and `ViT-B/16` at once (preferrable). `--dwt` switches to DWT (wavelets) generator instead of FFT. There are few methods, chosen by `--wave X`, e.g. `db2`, `db3`, `coif1`, `coif2`, etc. `--align XX` option is about composition (or sampling distribution, to be more precise): `uniform` is maybe the most adequate; `overscan` can make semi-seamless tileable textures. `--steps N` sets iterations count. 100-200 is enough for a starter; 500-1000 would elaborate it more thoroughly. `--samples N` sets amount of the image cuts (samples), processed at one step. With more samples you can set fewer iterations for similar result (and vice versa). 200/200 is a good guess. NB: GPU memory is mostly eaten by this count (not resolution)! `--aest X` enforces overall cuteness by employing [aesthetic loss](https://github.com/LAION-AI/aesthetic-predictor). try various values (may be negative). `--decay X` (compositional softness), `--colors X` (saturation) and `--contrast X` may be useful, especially for ResNet models (they tend to burn the colors). `--sharp X` may be useful to increase sharpness, if the image becomes "myopic" after increasing `decay`. it affects the other color parameters, better tweak them all together! Current defaults are `--decay 1.5 --colors 1.8 --contrast 1.1 --sharp 0`. `--transform X` applies some augmentations, usually enhancing result (but slower). there are few choices; `fast` seems optimal. `--optimizer` can be `adam`, `adamw`, `adam_custom` or `adamw_custom`. Custom options are noiser but stable; pure `adam` is softer, but may tend to colored blurring. `--invert` negates the whole criteria, if you fancy ch
Excerpt of 9,854 characters
Read on GitHub
164
tom white · New Zealand
15
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:3ba783fbc9ae66d9, topic:text-to-image, desc:text-to-image, readme:text-to-image
matched fp:3ba783fbc9ae66d9, topic:text-to-video
matched fp:3ba783fbc9ae66d9, topic:clip