Photonic Illusion V4 is my 3rd and possibly final illustrious checkpoint. This one has NoobAI merged, a few quality improvement LoRA's are baked in, it covers more prompting styles and I have added even more finetuned male data set training material and this one is much faster whereas my previous version I recommended 25-30 steps this one only needs 15 maybe 20 steps. I personally find 15 is all you need unless you have a lot of LoRA's in your prompt you may need the extra 5 steps to really flesh it out but a straight up prompt or just a couple LoRA's and 15 is perfect.
V3 is on here and V2 is exclusive to MageSpace. These are not new checkpoints just new to CivitAI.
Some basic recommended settings:
On A1111/ForgeUI I recommend Euler a + Automatic
Steps 15-20
CFG: 4-5
Size: 768 x 1024 or 768 x 1100 Upscaled 1.5 - 2x
Hires: Denoising Strength: 0.30-0.40 + Hires upscale: x2, Hires steps: 10-15 and for Hires Upscaler I recommend: 4x_foolhardy_Remacri (as my first choice) or R-ESRGAN 4x+ or R-ESRGAN 4x+ Anime6B
Like V3... For some of the sample images I have used some amazing prompts found right here on CivitAI, I wanted to test random amazing prompts I have seen from other models to see how well they performed on mine. So thank you to those that share their prompts!
Photonic Illusion V3 is my second true illustrious checkpoint with a larger finetuned male data set. V2 is exclusive to MageSpace and can be found here - https://www.mage.space/u/Stecfx
I have since renamed V1 to VR (Version Realistic) here https://civarchive.com/models/2346263/photonic-illusion-vr it was purely an experimental model which merged my first illustrious model that I didn't think was good enough to release on it's own along with my SDXL Finalé and Pony checkpoints.
V3 is an improvement over V2 with an even larger male focused data set added to the mix. This checkpoint can handle all female content but it should do even better with male focused content than most other illustrious checkpoints that should help make it a bit more unique amongst the crowd. It can still handle most general illustration styles as a well rounded all purpose checkpoint! I hope you enjoy and if you do please spread the word.
Some basic recommended settings:
On A1111/ForgeUI I recommend Euler a + Automatic
Steps 25-30
CFG: 4-5
Size: 768 x 1024 or 768 x 1100 Upscaled 1.5 - 2x
Hires: Denoising Strength: 0.30-0.40 + Hires upscale: x2, Hires steps: 10-15 and for Hires Upscaler I recommend: R-ESRGAN 4x+ or R-ESRGAN 4x+ Anime6B
If you find a better combination of settings again please share with the community!
For some of the sample images I have used some amazing prompts found right here on CivitAI, I wanted to test random amazing prompts I have seen from other models to see how well they performed on mine. So thank you to those that share their prompts!
If you like my models I gladly accept donated --> Coffee and Beer <-- your donations help me to keep motivated to continue making new checkpoints and LoRA's for all our enjoyment! To those that are generous enough to throw me a few dollars I thank you immensely! <3
Description
Photonic Illusion V4 is my 3rd and possibly final illustrious checkpoint. This one has NoobAI merged, a few quality improvement LoRA's are baked in, it covers more prompting styles and I have added even more finetuned male data set training material and this one is much faster whereas my previous version I recommended 25-30 steps this one only needs 15 maybe 20 steps. I personally find 15 is all you need unless you have a lot of LoRA's in your prompt you may need the extra 5 steps to really flesh it out but a straight up prompt or just a couple LoRA's and 15 is perfect.
FAQ
Comments (8)
Are these two models exactly the same as the ones that were previously published on the "tensor" platform?
Yes that's why I said in my comments "These are not new checkpoints just new to CivitAI." They were the last checkpoints to bring over here after I asked on here if people wanted me to bring them to CivitAI people said yes. My Photonic Illusion Chroma checkpoint is my newest model. At the moment I am taking a small break then deciding on my next model.
@StecFX OK, I'm looking forward to your new work!
Everything I make with 4 tends to come out blurry or distorted. I can't guide its linework at all. It seems to only work with the style it shipped with, and anything else really won't work. I'm always glad for diverse models, especially ones that remember male figures exist and that people like to draw them, so I hope that there can be some improvement.
V4 is a bit experimental and more geared towards my own personal preference, Does V3 work better for you at all for what you are trying to achieve? Appreciate the feedback.
V4's img is always covered with a blur? Don't know why. I'll download v3 to test.
Maybe that's vae's reason? problem disappeared.
@Gay_FlipFuck sorry you had issues but glad it sorted out. I always bake the vae in.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.