Photanima v2.1 showcase. Each image takes about 2 seconds to generate.

https://redd.it/1u04320
@rStableDiffusion

I think I'm finally feeling Ideogram 4...

Using this one nice workflow https://civitai.com/models/2679071/ideogram-fast-and-quality-ifaq-t2i-by-artgourieff and a LITTLE help of Claude.

https://preview.redd.it/hvr40r40m16h1.png?width=1072&format=png&auto=webp&s=6a1f747cc1d0c51fbc1fe9f4485015548186a756

https://preview.redd.it/4r0kxc40m16h1.png?width=1072&format=png&auto=webp&s=aca52fdf6d2f352b33681c67235b2a49ad47617a

https://preview.redd.it/ya7yms30m16h1.png?width=1920&format=png&auto=webp&s=466ec6897210ee2f0495566dfe0a7c98601ebca3

https://preview.redd.it/vagf0160m16h1.png?width=1920&format=png&auto=webp&s=401acea72ca17aa9722f4e1c485086a14d6b9a95



https://redd.it/1u04yjm
@rStableDiffusion

Ideogrammar — Ideogram 4 Prompt Editor

https://preview.redd.it/q6k6hw2fd16h1.png?width=5108&format=png&auto=webp&s=9a4c074e040604fe46e9b6f7ae09648556b3ced5


In case anyone is interested, I have made a visual Ideogram 4 prompt editor, with various features. Like generate the JSON by LLM with vision, vectorization to SVG and an integrated library/viewer. It can be used as standalone JSON generator, or it can be connected to a ComfyUI endpoint and generate directly.

To use it, pull the repo, set the endpoints and start generating.

https://redd.it/1u0429g
@rStableDiffusion

Lodestone is thinking about training ideogram! Prove him it's a good idea!
https://redd.it/1u09lxm
@rStableDiffusion

switched from Flux to Z-Image Turbo for character LoRAs, here's the honest comparison
https://redd.it/1u0a2dj
@rStableDiffusion

Ideogram 4 huge-res test: 8MP, 48 steps, 21 min on RTX 4090
https://imgur.com/a/eiCmDXf

https://redd.it/1u08l3e
@rStableDiffusion

Ideogram 4.0's Understanding of Characters and IP is Crazy for an Open Model

https://redd.it/1u0e1g0
@rStableDiffusion