renderartist
@Render_Artist
Followers
608
Following
5K
Media
364
Statuses
1K
Creating digital art out of bits, bytes, and electrons.
Joined August 2016
"bUt yoU CaN maKE tHiS wITh aNY biG AI MoDel" Actually, you can’t (yet). The outputs will be inconsistent, stylistically messy, and mostly unusable. That’s the tiring part of @X AI these days... people have become lazy, thinking everything is just 1 click away. It’s not.
@emmanuel_2m Also just give any big AI image gen an image of a buildilng and say "Generate an image of this building for me but in an isometric style with a white background". . Stop trying to con people into thinking they need a middle man(software) like you. We don't.
20
23
357
I gave Opus 4.1 access to a pen plotter- And asked him to draw several self-portraits. Here are the results:
123
208
2K
Had a dream last night that everyone went nuts watching viral AI videos, Sora kinda nailed it aside from the gimped quality all the videos seem to have now. 🫣
0
0
0
Karen Crashouts in Sora 2, mostly wanted to see how consistent I could get a theme to work across multiple clips with some scripted dialogue. Had fun with these. 😂 @openai
0
0
0
More @OpenAI Sora 2 stuff from last night. Trying to direct a lot of what’s happening in the clips with JSON prompt structures but sometimes physics fall apart.
1
0
1
Published Saturday Morning WAN high and low noise LoRAs on @HelloCivitai and @huggingface today. Trained with WAN 2.2 14B T2V, lots of fun. Music with @SunoMusic ✨
1
0
1
Eye gaze fully fixed in Comfy now for Wan Animate. All you need to do is use the pre processing nodes I linked in the first comment below for your pose / face crop. Tested this against Wan's Native source code and it is indeed 1:1 functioning now this was 300+ frames
22
33
411
The agent mode in @cursor_ai has gotten so good, you can just vibe code your own UI for any @replicate API endpoint in under an hour now, what the hell. I used Seedream 4 via this custom React/Typescript UI to generate some of my training data for my latest LoRAs. 🙌🏼
0
0
7
Someone tried one of the images with WAN 2.2 Animate https://t.co/wEQPflhxBs 🔥
0
1
1
0
0
0
1
0
0
Published my Saturday Morning Qwen LoRA on @huggingface and @HelloCivitai - This variant of the model is more expressive and a closer match to the training data.
2
0
0
Saturday Morning Flux is available to download from @huggingface - Grab it here https://t.co/baGCdC48ZX
0
0
0
Uploaded Technically Color Qwen to @HelloCivitai and @huggingface - This is my first time training with Qwen, interesting model. Still trying to find the best strategy for inference, I've attached my workflows to the images in the respective image galleries. ✨
1
0
1