If you do like this work, consider buying me a coffee :)
Updated - Mid Year: Updated with a lora created from a frest dataset merged with ICBINP Final. Should give decent results, but does have the standard AI issues (cleft chin bias etc..)
Recommended settings: DPM++ 3M SDE Karras or DPM++ 2M Karras, 20-30 steps, 2.5-5 CFG (or use Dynamic Thresholding), happiest at 640x960 with a hires fix on top, but if you are happy to hunt through seeds to avoid head duplicates, 768x1152 works quite well also.
Following on from Gorilla With A Brick, I've merged in 10 more photorealistic models at various weights, and some more noise offset to create something that when prompted for photorealism will make you go "I Can't Believe It's Not Photography". It will happily create CGI characters and awesome landscapes as well.
As always, pruned to fp16, the VAE is baked in (SD-v2 840000 - ignore the VAE requirement above as the system is borked!), and the man in the suit can be used as the calibration prompt on any UI!
Description
MBW model merges with Photon, Serenity, and Realistic Vision 4, and ICBINP Final, then lora merges of RCNZ_HQR, LowRA, Polyhedron's Skin lora, and add_saturation (negative merge)
FAQ
Comments (33)
Using Automatic1111. What do you mean by calibration prompt?
As in it's super simple and it doesn't matter if you are using a basic python script calling the huggingface diffusers pipeline or a1111/ComfyUI/Easy Diffusion etc, you will be able to easily get the same result (assuming the model downloaded correctly).
Dude! Outstanding new model, well done, each time you outdo yourself again! Thankyou for all the work you've put in and I hope we see you in SDXL
great model... Reminds me of work lol 😂.
FA_ICBINP_FINAL_CLIENT-EDIT2_LAST-EDIT3_REAL-FINAL4.psd
123456
78910,11,12
11,12!??
It isn't safe to type your password out like that.
Great. Now I'll be humming that song for the rest of the day. Thanks a bunch.
What is the different between FINAL and RELAPSE? I can I replace Final with Relapse and not lose anything?
They are different models. I feel like FINAL gives a smoother, more airbrushed output, but there are comparisons on the poster images for SECO and AFTERBURN that have FINAL and RELAPSE next to each other if you want to see for yourself
This is inane! the level of quality in this WOW
Tutorial for this model ICBINP Photography realistic easy and fast
Nice work :) For those that don't want to click without knowing, its a link to a video that shows how to use Google Colab to run this model using a1111 if you don't have a GPU
How can contact with you?
I'm residentchiefnz on discord, or post your details here and I can contact you
What is the difference between FINAL and SECO?
More models in the merge, and better use of the "fixing" LoRAs. There are comparisons on the SECO model card if you want to see them side by side
Really great for drawing realistic boobs, but nipples are always bad
pretty nice results, but seems a bit prone to draw women sexy and busty. the simple prompt of "X in the park watching the doves fly" yields very different results if X is a man or a woman
I noticed A LOT of models have this 'feature' even when using lots of negative prompts, and this is for regular models advertised as SFW. It can be very time consuming... I'm actually still looking for a model that doesn't do this. Have you found any?
@usevalue try realistic vision 5.1
@usevalue LOL, I've had some models that work perfectly fine with "woman with brown hair" and be SFW, but as soon as I change brown to blonde... the women are all nude seed after seed.
I'm really sorry to report this is unseibule
it's genarating me very badly done sceatches or drawings i want photos like these previews
please could you kindly tell me what is going wrong?
I loved this modal on Catbird AI
but it's not working in EasyDefusion
thabks in advance
Hi there @StableDEFERAZER , to me it sounds like something might be astray with your Easy Diffusion setup. See if you can replicate the image like https://civitai.com/images/1635474?modelVersionId=121557&prioritizedUserIds=65239&period=AllTime&sort=Most+Reactions&limit=20 in the system (or find the image for the same prompt in the version of ICBINP you have). If you can create that one, then the issue is with EasyDiffusion not able to process in the same way a1111 or ComfyUI can (i.e prompts are getting cut off due to still having a token limit or any of several other possibilities). If you can't replicate that one, then I'd be looking at reinstalling EasyDiffusion or trying Automatic1111
Do you have Discord? How can I chat directly with you?
Yup, Im residentchiefnz on discord
Hi, Is it allowed to use your model to be merge with other one and then post on Civitai. Please let me know. Thanks
More than welcome to!
Thanks a lot
My photos are coming out with some weird blemishes, like weird decolorations in the skin, it's not a VAE error, it's something else... Any idea what might be causing it?
CFG too high?
@residentchiefnz It was trouble with the seco model, and I think it was down to the resolutions I was using not working great. Using final version and no issues!
@Malessar Glad you got something working :)
Details
Files
icbinpICantBelieveIts_seco.safetensors
Mirrors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
ICBINP v8.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors
icbinpICantBelieveIts_seco.safetensors