Workflow Venice AI Workflows and Models (DON'T POST REAL PEOPLE!!)

thanks for the info, I've tried that but my nsfw fails, that's why I didn't think that was the idea, does your combine images nsfw prompts work?
EDIT: now one passed, maybe combine has a chance to fail if nsfw?
 
Qwen Image Edit Uncensored is the primary one since it doesn't directly cost credits. If I have credits to spare, SeaDream v5 Pro is really good. The only drawback is that it's 12c and it takes longer to generate.
 
seedream 5.0 is crazy good (combine)
"reproduce the second image with the girl in the first image" always work
any prompt share ?
 
good quality from seedream 5.0 (on venice), getting close to grok HQ, lacks more prompt adherence, not as realistic yet
 
tits still look very odd and plastic to me. yeah it is a good model but i don't think it is in Grok's NSFW level.
ByteDance still don't use porn when training their data probably. All of their models are weak for porn compared to Grok.
 
yes, not as realistic, but it is the first model I would consider using for I2I after grok HQ, tits are hit or miss, it makes them too perfect usually, I got some better results by adding "saggy", "swollen", "natural realistic"
 
im at a loss, anyone know how to make audio files work in seedance 2.0 fast r2v? i used m4a, i've also used mp3 and i still keep getting this same error:
The model returned the following error: Video generation parameters are invalid. Please adjust your settings and try again.
 
It should take mp3 or wav. Just make sure the audio files are less than 15s and less than 15MB. I've had Seedance complain about a video ref that was just slightly over 15s and had to trim it with ffmpeg.
 
Truly the best thing I can recommend is using agentic chat kimi (free) ask it to learn everything about prompting image edit generation for that model then tell it that it’s a professional master class nsfw image prompt director, then tell it what you want.
 
Does anyone else have the problem of their chats hanging and not loading? I really don't want to clear my cache and wipe all my data. Is there a workaround?
 
Here's my full workflow from text to image generation to character sheet.
I like Grok because it's really good at consistency, but you can start with Chroma or anything else you like. If you do use a censored model like Grok, you need an uncensored model for the final edit.
Step 1 - generate an image - GROK Image High Quality SOTA
Prompt
A young beautiful asian woman with large breasts, exceptionally wide hips proportions, thick legs, large buttocks, delicate skin details including cellulite, dimples and stretch marks, barefoot, posing in a front-facing quarter view.

Step 2 - generate a character sheet - GROK Image High Quality SOTA
Prompt
16:9 image, Create a triptych composition showing three poses of the same woman in a real room setting with natural sunlight streaming through a window onto the floor. Each panel is separated by clean white border lines. Left panel: 3/4 angle view cropped below mid-thigh, hand on hip, confident pose. Center panel: back view showing full body from behind, elegant posture. Right panel: front-facing full body view with hands on hips. Natural window lighting casting soft shadows on the floor, photorealistic interior setting with visible floor and wall details, high quality photography.

Step 3 - use the character sheeet to generate your final image - Seedream V5 Pro
Prompt
use the woman provided in this reference sheet, she is sitting on a folding chair on balcony during golden hour, she's completely nude, her legs are spread and she has one hand on her labia, the other hand is squeezing one of her nipples

Final Image
 
Anyone else disappointed with the new upscale/enhance model? The old one did a good job of smoothing out the the randomly ugly noise you sometimes get with AI generated images. The new model doesn't seem to have any creativity options at all. 😥
 
I can't upload images to do I2I prompts on the Seedream V5 Pro model or the Qwen Image 2.0 Pro model on Venice. It keeps saying the model doesn't support style references anyone else noticing this.
 
You are on the wrong mode, there is the edit mode next to image mode, that always editing with references
 
This is the discussion and workflow (including prompts) thread, too many models on Venice tho. Most people are lost and need help with non prompt related issues.
 
Some observations:
Use the right nudifier for the job. Qwen Image 2 pro has better breasts and skin detail, but colors are desaturated. Meanwhile, Seedream 5 pro has better pussy and colors, but details aren't as good
 
You must log in or register to reply here.
Back
Top