Sign In

PixelWave

2.7k
38.7k
877
Type
Checkpoint Trained
Stats
14,707
Reviews
Published
Oct 25, 2024
Base Model
Flux.1 D
Training
Steps: 382,000
Hash
AutoV2
A34D7B4F23
The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.
IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.
Please support civitai and creators by disabling adblock

PixelWave Flux.1-dev 03 fine tuned!

Safetensor Files: 💾BF16 💾FP8 💾NF4

GGUF Files: 💾Q8_0 🤗Q6_K 💾Q4_K_M

Links to 🤗VAE 🤗T5xxl 🤗CLIP L

The 'diffusers' files are actually the Q8_0 and Q4_K_M GGUF versions. GGUF files also available on huggingface.

I fine tuned version 03 from base FLUX.1-dev for over 5 weeks on my 4090. It is able to do different art styles, photography, and anime. Trick I discovered to help with LoRAs.

I used dpmpp 2m sgm uniform 30 steps for the showcase images. If you want a neater/cleaner output, try increasing the guidance. Also mentioning a style can help, so the model doesn't have to guess.

I also recommend try adding the upscale latent by node, and scale the latent by 1.5, e.g. generating an image that is 1536x1536 instead of 1024x1024.

PixelWave Flux.1-schnell 03

Safetensor Files: 💾FP8 💾NF4

GGUF Files: go to huggingface

I used dpmpp 2m sgm uniform 8 steps for the showcase images.

You can start with 4 steps, but there are less errors with anatomy if you run with more steps.

PixelWave Flux.1-dev 02

Safetensor Files: 💾BF16 💾FP8

GGUF Files: 💾Q8_0 🤗Q6_K 💾Q4_K_M

Version 02 has greatly improved black and dark images, and more reliable outputs with fewer issues with hands.

I recommend using dpmpp_2s_ancestral, beta, 14 steps. Or euler, simple, 20 steps.

Comfyui-GGUF Nodes

PixelWave 11 SDXL. A general purpose fine tuned model. Great for art and photo styles.

I use 20 steps, DPM++ SDE, CFG 4 to 6 or 40 steps, 2M SDE Karras

Accelerated Version - 5+ Steps, DPM++ SDE Karras, 2.5 CFG

PAG Recommended⚡Recommend 1.5 Scale, with CFG 3. Link to workflow

🔗Link to Expanded Gallery 🖼️

Link to prompting guide.⭐ You don't need to use 'quality' terms such as 4K, 8K, masterpiece, high def, high quality, etc. Unless you want it, I recommend not using words such as 'vibrant, intense, bright, high contrast, neon, dramatic' for photographic styles if you a wanting a more natural look. This can cause images to look 'overcooked', but it's just the CLIP following your prompt. 🙂 If you do want vibrant, neon photos PixelWave will provide!

The focus for version 10 was to train the CLIP models, which improves the reliability, ensures you can produce a wide variety of styles, and better at following prompts.

Thanks to my friends who helped test: masslevel, blink, socalguitarist, klinter, wizard whitebeard.

Guide: Upscaling Prompts with LM Studio and Mikey Nodes

Guide: Add more details to your image using the skip step method

No need for the refiner model.

This model is not a mix of other models.

I also created Mikey Nodes which contains a lot of useful nodes. You can install it through comfy manager.

🌟 Support My Work 🌟

If you've enjoyed my work on PixelWave and want to show your support, I have a slightly different request. Instead of monetary contributions, I ask for kindness and positivity:

  • Be Kind to Each Other and Yourself: A little kindness goes a long way. Treat yourself and those around you with compassion and understanding.

  • Donate to Your Favorite Charity: If you're able, consider donating to a charity that resonates with you. Your generosity can make a real difference in someone's life.

  • Call Your Parents (or a Loved One): Take a moment to reach out to your parents, a family member, or a close friend. A simple call can brighten someone's day.

  • Tell Your Partner or a Friend You Love Them: Remind the special people in your life how much they mean to you.

Your actions, big or small, can create ripples of positivity. Let's make the world a better place, one kind gesture at a time. Thank you for your support! I hope you have a wonderful day! 💖

Suggested Resources

Discussion

vitherenegade's Avatar
vitherenegade

I don't know if this is a bug but each time i write large breast in prompt or large bust every single gen has nipples showing how can i prevent it?

sevenof9247's Avatar
sevenof9247
SDXL Image Contest Participant

link in your notes ???

what is what ???

Full Model nf4 (6.46 GB)

Verified: a month ago

GGUF

Full Model fp8 (11.85 GB)

Verified: a month ago

GGUF

Full Model bf16 (22.17 GB)

Verified: 2 months ago

SafeTensor

Full Model fp8 (11.08 GB)

Verified: 2 months ago

SafeTensor

Full Model nf4 (5.72 GB)

Verified: a month ago

SafeTensor

MM
mmdd2543

Hi! I have a suggestion to enhance the model's performance in the next version. I've noticed that the model often struggles to accurately generate coniferous trees when I try to upscale images containing them in img2img workflows, often producing trees with regular leaves instead, even though I have the words "coniferous" and "pine needles" in my positive prompt. However, I have noticed that the model does create proper coniferous trees in txt2img workflows. Am I doing something wrong? Interestingly, in my tests I've observed this issue to be less common in most of the popular realistic Stable Diffusion 1.5 models, which are far less complex than PixelWave and Flux. I hope this feedback is helpful. Thanks for considering it! :) BTW, the model is FANTASTIC nonetheless. I want to express my gratitude for the high-quality work. It's clear to me that a lot of effort went into it and it really shows by all the beautiful creations showcased in the gallery.

Pechista's Avatar
Pechista

Any plan to add the Hyper-8-steps model version?

Janet's Avatar
Janet

how well does it do males and various ages?

why not train over the dedistilled or libreflux? Wouldn't results be better?

MasterArt's Avatar
MasterArt

Very good! An update would be great too.

PO
Pondloso

This is the best flux model for me in term of artistic.

But is had very serious down side....clone face and body shape.

It love to always put out similar face and same body shape no matter how i prompt(with same main object).

and if i want multiple woman in one image they face and body just clone of each other.

If you want to fine tune more i would love you to put a lot more face and diverse body type into training data.

This is really great flux model but this clone face it very bad to use.

See 15 more hidden comments