🚀 NEWEST RELEASE — JUNE 2026
🔥 FameGrid Spice V2
The Biggest Quality Jump Yet
After extensive training, dataset refinement, and countless iterations, FameGrid Spice V2 represents the largest leap forward we've achieved so far.
This release focuses on the areas that matter most in real-world image generation: realism, skin quality, pose quality, prompt understanding, image consistency, and overall visual fidelity.
Whether you're creating influencers, virtual creators, social media content, commercial campaigns, lifestyle photography, or premium creator content, Spice V2 delivers stronger results with less effort and greater consistency.
✨ What's New
✅ Improved Skin Quality
More realistic texture, better detail retention, and significantly more natural-looking skin.
✅ Improved NSFW Capabilities
Better anatomy handling, stronger consistency, and improved prompt adherence.
✅ Improved Posing
More natural body language, cleaner hand placement, and stronger compositions across a wider variety of scenes.
✅ Improved Prompt Following
More accurate interpretation of prompts with better adherence to complex instructions.
✅ Wider Ethnicity Representation
Improved diversity across generations with stronger representation of different ethnic backgrounds and facial features.
✅ Expanded Body Type Variety
More realistic and diverse body shapes, proportions, and physiques for greater creative flexibility.
✅ Improved Male Generation Quality
More attractive, realistic, and consistent male subjects with improved facial structure, styling, and overall appearance.
✅ Reduced Image Artifacts
Cleaner generations with fewer distortions, visual anomalies, and unwanted artifacts.
⚙️ Recommended Settings
Sampler
sa_diag_implicit / qin_zhang_2s
Scheduler
beta57
Steps
9 (20 steps on the ClownShark BetaSamplingScheduler)
CFG
1.0
ETA
0.50
Denoise
1.0
Sampler Mode
resample
Bongmath
true
These settings were used throughout development and consistently produced the strongest results.
💎 Free Premium FameGrid Spice Workflow Package
Take FameGrid further with our premium workflow collection.
Included
✅ Image-to-Prompt (Single Image)
✅ Image-to-Prompt (Folder Processing)
✅ Custom Grok Prompt LLM Node
Available exclusively through the UltraMuse Discord.
👉 Join Here: https://discord.gg/4zbGm5j6jW
🎬 Introducing Creator Studio AI
The AI Influencer Operating System
Creator Studio is a complete content production platform built for AI creators, virtual influencers, agencies, and power users.
Unlike most creator platforms, Creator Studio isn't a closed ecosystem.
It works with your existing workflows.
Whether you're creating a single virtual creator or managing an entire roster of AI influencers, Creator Studio gives you the tools needed to create, manage, and scale content production from one platform.
https://ultramuse.art/creatorstudio/
🚀 Everything You Need In One Place
We tested over 100+ models / platforms and placed the best all in one easy to use and powerful UI.
Creator Studio combines powerful AI tools into a single creator-focused production environment.
No more jumping between websites, prompt files, spreadsheets, Discord servers, and disconnected workflows.
Build your entire creator pipeline in one place.
🖼 Unrestricted Image Generation & Editing
Generate high-quality images using your preferred models, styles, and workflows.
Create content for:
✨ TikTok
✨ Marketing Campaigns
✨ Virtual Influencers
✨ Lifestyle Content
✨ Product Promotions
✨ Creator Brands
🎥 Unrestricted Video Generation
Transform images and concepts into engaging video content.
Create short-form content, social media videos, promotional assets, and creator-focused visual storytelling.
🤖 AI Agents
Automate repetitive creator tasks and streamline your workflow.
AI Agents help reduce manual work so you can spend more time creating and less time managing production.
📸 Character Management
Maintain consistency across your AI creators.
Manage characters, appearances, styles, references, and creator assets without constantly rebuilding prompts from scratch.
Perfect for virtual influencers and long-term creator projects.
⚙ Workflow Automation
Build repeatable production pipelines that improve consistency and save time.
Reduce trial and error while creating content at scale.
⚡ Bring Your Own ComfyUI Workflows
One of Creator Studio's most powerful features is native support for custom ComfyUI workflows.
Import Any Workflow
Creator Studio allows you to:
✅ Import existing ComfyUI workflows
✅ Use custom nodes
✅ Use custom checkpoints
✅ Use custom LoRAs
✅ Use custom models
✅ Use your own generation pipelines
If you've already invested time building workflows, you don't have to start over.
Bring them directly into Creator Studio and use them alongside all of the platform's creator-focused tools.
🛠 Built For Advanced Users
Most platforms force you into their workflow.
Creator Studio lets you keep yours.
Whether your workflow includes:
Flux
Qwen
Z-Image
SDXL
WAN
Custom Fine-Tunes
Character LoRAs
Face Swap Pipelines
Upscaling Pipelines
Video Generation Workflows
Creator Studio is designed to work with them.
📈 Built For Scale
Creator Studio was designed for creators producing content professionally.
Whether you're running:
✨ One AI Influencer
✨ Multiple AI Creators
✨ An Agency
✨ A Creator Team
✨ A Virtual Talent Network
The platform helps you move from individual generations to a repeatable content production system.
🌐 Creator Studio AI
Your AI Influencer Operating System
✅ Prompt Studio
✅ Image Generation
✅ Video Generation
✅ AI Agents
✅ Character Management
✅ Workflow Automation
✅ Native ComfyUI Workflow Support
Create faster. Stay consistent. Scale smarter.
👉 https://ultramuse.art/creatorstudio/
💬 Join The Community
The UltraMuse Discord is where you'll find:
✅ Official Workflows
✅ Premium Workflow Updates
✅ Additional Checkpoints
✅ Tutorials
✅ Prompt Resources
✅ Troubleshooting Help
✅ Experimental Releases
✅ Community Showcases
✅ Early Access Releases
✅ Creator Studio Updates
👉 https://discord.gg/4zbGm5j6jW
📚 Legacy Models & Previous Releases
Looking for previous FameGrid releases?
Many of our earlier models remain excellent options depending on your workflow, preferred model family, and desired aesthetic.
🚀 FameGrid Revolution
Our previous flagship release built around the Z-Image ecosystem.
Features
✅ Compatible with Z-Image Base
✅ Compatible with Z-Image Turbo
✅ Excellent Prompt Adherence
✅ Strong Social Media Aesthetic
✅ High-Quality Influencer Photography
Resources
👉 Model Page
https://civarchive.com/models/2431502/famegrid-revolution-hybrid-workflow-zib-zit
👉 Official Hybrid Workflow
🎨 Revolution Variants
FameGrid Revolution Bold
Stronger baked-in influencer styling with a polished Instagram aesthetic.
FameGrid Revolution More Bold
Maximum social media styling and stronger beauty influence.
FameGrid Revolution ExtraReal
Cleaner realism, more believable subjects, and authentic lifestyle photography.
FameGrid Revolution UltraReal
Maximum realism, cleaner environments, stronger adherence, and fewer artifacts.
Additional variants and checkpoints are available through Discord.
🆕 FameGrid Qwen Series
The FameGrid Qwen releases introduced improved skin realism and more believable subjects.
Trigger Words
IGMODEL
Main trigger for influencer-style realism.
rlskn
Enhanced skin texture, natural imperfections, and more believable subjects.
Example:
iPhone photo of an IGMODEL woman with rlskn, detailed skin texture, natural imperfections...
⚙ Recommended Qwen Settings
Model Flow
2
Sampler
res_2s
Scheduler
simple
CFG
1–3 (1 recommended)
LoRA Weight
1.0–2.0
(1.0 recommended. Increase slightly if realism begins to drop.)
🎯 Best Uses
✨ Influencer Photography
✨ Lifestyle Content
✨ Virtual Creators
✨ Product Photography
✨ Marketing Campaigns
✨ Social Media Content
✨ Character Creation
🌐 Important Links
Creator Studio
https://ultramuse.art/creatorstudio/
UltraMuse Website
UltraMuse Discord
FameGrid Revolution
https://civarchive.com/models/2431502/famegrid-revolution-hybrid-workflow-zib-zit
❤️ About FameGrid
FameGrid was built around one goal:
Create modern social-media-quality imagery that feels believable, natural, and visually compelling.
The focus has always been:
Natural posing
Realistic skin
Clean compositions
Modern lighting
Strong prompt adherence
Professional creator aesthetics
Whether you're building influencer content, virtual creators, marketing assets, product photography, or commercial campaigns, FameGrid is designed to help you produce better results with less effort and greater consistency.
Description
FAQ
Comments (59)
Huh, how come the cfg used is 7?
That's wrong. cfg is 1 for all the images.
@UltraMuse I was just looking at your images examples that has 20 steps and 7 that's why lol, thank you that makes sense,will do some test,so Far it seems to make the skin with less pores or textures and some random noise on the image,will post some
@Neon_signs Also, try 8 steps. 10+ smooths the skin out.
The ZIT version of this lora seems to make skin/the whole image more plastic when used at higher weights, most z-image loras seem to cause this, possibly as we are still stuck with just the Turbo Model for now.
I thought it was just me lol
true. training a Lokr factor 4 instead of regular LoRAs did improve my results quite a bit, but it's still not perfect though. (ai-toolkit, de-turbo model)
Yeah, anything past 12 steps also gets mushy. Waiting on the base to save the day.
UltraMuse, incredible result!
What was the dataset size and training parameters you used?
I believe it was 3K and 45000 steps
Why discord? :(
It's easier/quicker to offer support there.
Pretty good, but struggle to generate Asian women (at least with the Z-image LoRA)
FANTASTIC WORK!!! Amazinggg realism!! This is a next level of image generation! Oo
The sample images you shared are truly incredible. I downloaded and ran Lora, but the outputs aren't even close to the ones you shared; 90% of them are poor, and the results are the same in the recommended workflow. Even if I occasionally get good results, most of them are poor. Even with the same prompt and resolution, they are far from the sample outputs. My goal is not to criticize, but where am I going wrong?
Hi there, sorry to hear. Are you are using the workflow from the Discord?
@UltraMuse Yes, everything is the same.
@goldennyks76 Are you running any type of optimizers? join the discord and see if we can figure it out. https://discord.gg/aNnuuhJT
@goldennyks76 mans begging you to join the discord lmao did you figure out a solution?
@Thatonecreator Discord is banned in my country.
@UltraMuse Just release the workflow here
Fuq yo discord, bot move
Makes more sense to upload workflows here
The workflows are on Discord for a reason. Many people dismiss the lora without actually trying or some wrong implementation so it's for support.
@UltraMuse I'll upload the workflows here for you since you're doing shady shit with that forced website login to even join the discord, you can easily give support on Civit too... As I said, BOT MOVE
@Cameronie I'm pretty sure the 100s of hours and thousands of dollars I spend on creating the models pale in comparison to the 2 minutes it takes to sign up...
@UltraMuse You make it sketchy nonetheless! Discord is one thing but forcing users to sign into this junk is shady: UltraMuse | Coming Soon
Nobody cares about the money you invested, the fact you mentioned that makes it even more weird. Begging for joins at that point
@Cameronie 200 members didn't seem to think so. It's using common practices for verification via their own API. Also, no one is forcing you to join, you can free to use your own workflow.
@UltraMuse I did join, snagged the workflow and left. I'll upload it for you. As I said.
@Cameronie You might want to read civit TOS before doing that lol :)
@UltraMuse They'll understand based on your incompetence and directing people to sketchy websites. I'll make sure it stays up either way
@Cameronie 🤦🏽♂️
@Cameronie Can you send me the workflow
@Thatonecreator Nah, workflow was trash. There are better options out there, exactly why I didn't waste anymore time on this peasant
@Cameronie yea nah it was
Are those images clearly generated from this no addition?
not possible, i cant get even one image good like this quality
Yes, the only other difference is that I used a prompting tool I made that formats prompts into a way that the lora likes.
https://chatgpt.com/g/g-691d0029547881919fdf201b163ac788-famegrid-prompt-expander
Also join the Discord theres a growing community there and we share results.
@markhassain3712 Join the Discord, there are plenty people getting good results. Also we offer support there.
@UltraMuse okay
@UltraMuse where can i find your discord
@nitza12345job682 https://discord.gg/4zbGm5j6jW
@nitza12345job682 did you find the solution?
@markhassain3712 did you find the solution
@Thatonecreator well its 20 steps
@markhassain3712 20 steps for z image??????????????
@Thatonecreator yes
@markhassain3712 thats crazy, i'll will try that and get back to you
the discord bot not works
Should be working now.
Can you just be convenient and put the workflow here or at least send it to me?
"Join servers for you" permission just to verify in the discord with your sketchy bot? no thanks. why not just post the workflow here like a normal person?
You can login to discord first then go to the page if you don't want to login there.
I second this. Why not just release the workflow here? This is so sketchy
@tastefulaiDA How so lol. Its just a simple Oauth that takes you to my website for a survey of what tools you use, then auto-joins the server. Nothing more than that. You'd think someone would say something by now if otherwise..
@UltraMuse Release the workflow here or delete your Lora or change the sample images. Your Lora is entirely misleading.
@coolstrad I second this. Nobody can reproduce the sample images and the OP wants users to join their Discord instead of just publishing the workflow here
@UltraMuse Not publishing the workflow on Civitai and offering NO help at all to reproduce your images, except for "join my discord to download exclusive workflow" does not fit into the community and is misleading. Remove your Lora if you are not willing to offer information on how to reproduce your images.
@tastefulaiDA Do you not see the gallery? There's plenty of people getting the same results... The point of Discord is the help..but if you look at the other comments I have been answering questions to those asking politely. You haven't, just been complaining lol.
@UltraMuse You're clearly ignoring all the comments. I am not complaining about you helping in Discord. We, several users, are telling you to publish your workflow on Civitai instead of hiding it and making people force-join your Discord. Nobody wants to join your Discord.
Details
Files
Famegrid_Z_Image_V0.5_Beta.safetensors
Mirrors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
FamegridZImageV0.5Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Famegrid_Z_Image_V0.5_Beta.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.


















