4B model in the 9B category. Can you make a 9B version? EDIT: HE FIXED

So what does this achieve? Is base with this lora much better than using the turbo checkpoints in general?

Does that mean it can be used as CFG 1 in the base model?

Did some testing of 9B rank 128 version - it works very well indeed, thanks for making it ! i also noticed what 9B base + lora actually runs faster than regular distilled 9B even with negative prompt and cfg at 5, but not just that - it looks like it is "smarter" too when it comes to stuff like landscapes\cities\sci-fi etc - regular 9B is definitely not as good in that department - maybe it got a little too overdistilled towards humans generation because bfl tried to make z-image killer (which they totally did))) but overcompensated with that a little, trying to have a good pr this time by appeasing people who howled Flux 2 dev was bad and useless because it took too long generating their only interest in all this ai stuff - them nekkid chicks ofc ))))

This seems to work pretty good! Thanks! If I set the LoRA strength to ~0.5, I can reduce the steps from 40–50 to 10–20, with lower CFG (usually 2.0 - 2.5). It gives me the best of both worlds - the variety and tack-sharp image quality of Base + the aesthetic tuning of the distilled model, all at much quicker generation times.

This is a huge help to CFG

How much control do you want? YES...

Do 9b klein LORA's work in combination with this LORA?

"This was created by subtracting klein-base from klein and creating a LoRA from it."

Can "z image base" also be used to create Turbo Lora?

What did you use to extract it? I tried the easy way using ComfyUI-FluxTrainer, but just got noise. Your result is pretty close to the original

An incredible LoRA, it made my life 500% easier.

🙌💖🙌

The checkpoint I was using insisted on maintaining a saturated color, regardless of the settings, but your LoRA not only sped up image generation, but also gave a magnificent touch to the model's color tones.

Thank you for the beautiful work you did.

any proven good lora weight/cfg/step count combination suggestions?

Any reason to use the rank 64 other than people with low VRAM?

Just awesome thank you! it works very good with sampler LCM/simple or beta57 even with CFG 3.0. Recommend for you guys to use Flux.2 KLEIN enchancer custom node.

Nailed it. Took it apart, rebuilt it properly, wired some of the logic from Qwen — now every run just like magic. I will share my workflow after I clean all private stuff. but its gold

Can we definitely use negative prompts with this BFL says that Klein doesn't support them

It would be good to have a short description how you suggest to use it properly. Number of step counts, samplers, difference between ranks, CFG, strength etc.

Just want to say: extraordinary! Negative prompt AND fast generation!? And it works great. Thank you so much! Good work!

What settings are people using for this? At full strength, and at partial strength. There's no info in this description.

probably a noob question but whats the difference between the 9b 64 and 9b 128?

My Lora Manager can’t detect the Klein 4B/9B Base to Turbo, it detects other Loras in my Lora folder fine but not this one, anyone else having the same issue?

noob question, why and when do use this? I tried plugging it in to a workflow and the results looks like im on lsd lol

I've noticed that using this Lora seems to reduce editing strength – sometimes the image shows no change at all compared to not using it.
I noticed that when CFG is set to 1, the first step follows the prompt correctly, but from the second step onward, it reverts back to the original image. Only when I increase CFG to 1.2 does the editing work properly — though at the cost of roughly double the computation time.
Just wanted to flag this. Thanks!

“Hi there, could you briefly explain the process behind creating this type of acceleration LoRA?”

So is it still 4 Steps, because I think thats the biggest issue with Klein which leads to the anatomy issues, if it 12 or even 8 steps and I think it would be fine.

honestly for image editing thats where Flux Klein Base 9B shines it's so stable that with this workflow https://comfy.org/workflows/templates_doc_workbox_klein_9b_image_extend-a8d874e28508/ and adding the power lora loader to it replacing the prompt with "Remove the #00FF00 areas in image 1. Fill in the #00FF00 areas with a: ... [describe what you want]" you can take a head shot and extend it out to a full body image and "#00FF00" is the Hex code for RGB 0,255,0 aka pure green. You will want to set the Consistency Lora:0.5 and Flux2 Klein Turbo Lora:0.5 using 12 steps and keeping the CFG at 1, keep the Euler Sampler and if you also add in a draw mask on on directly between the "Load Image" node and the first "ImageScaleToTotalPixels" node again obviously setting the "Draw Mask On Image" to color: 0, 255, 0 and using "Realism_Engine_V2" lora you can extend a any headshot into an NSFW image while retaining the face, reason for adding another "Draw Mask On Image" node is to cover the clothes and if necessary covering the arms in a mask and shoulder blades so you can extend out the image with the arms in whatever position you prompt rather than being stuck with the position the top of the arms and sholderblades would suggest, also it's to get rid of stubborn background elements. You could automate the masking and for the background at least replacing the positive reference latent with the custom node "Reference Latent+" which comes with an auto masker—google Reference Latent plus—you can mask off the background automagically, and you could use a person BOX and SEG to automatically mask the person but reliability is a "bit" shaky detecting a person in a head shot and if it does work it wont mask the persons arms and sholderblades, you could use a face seg which would be more reliable, but I don't know how that would turn out,

also a great tool for processing manga panels before turning them into videos, but you will probably want to turn off padding and just manually draw over any elements you don't want in the final image.

Also yes Flux Klein has a limited understanding of both Pantone and Hex color code values, as the data set is a cut down version of Flux2 Dev which has a more full understanding, but #00FF00 is so unambiguous it knows what that looks like.

Hi could you tell me the general purpose/concept/use case of this, I understand flux has 4b and 9b base and distilled so 4 variants for the lesser versions of flux. Distilled is the 4 steps while the bases are the 20 steps. So all of those should theoretically have some niche or use case, then this is something converting on to another, but how would you explain to a 6-year-old, why? Thanks!