Z-Image version is available!
Nova Reality
Nova Reality is realistic SDXL checkpoint with great feels of details
All images on example were created with diffusers with custom png tags
Rules
You cannot use the generated images for commercial use if it's not edited (or just turning it to black and white)
You can share images without any restriction if you don't monetize it
Advertising this model to outside is always welcome
Recommend Settings
Sampler: Euler a
Steps: 20~30
Clip Skip: 1-2
Denoising Strength: 0.65 - 0.8
(Pony)
CFG Scale: 5~7
Prompt: score_9, score_8_up, score_7_up, realistic, photo, detailed face, detailed skin, BREAK
Negative Prompts: score_6, score_5, score_4, score_6_up, score_5_up, score_4_up, source_anime, source_furry, source_pony, source_cartoon, blurry, jpeg artifacts, normal quality, worst quality, low quality, missing fingers, extra digits, fewer digits, bad eye
(Illustrious)
CFG Scale: 4-6
Prompt: masterpiece, best quality, amazing quality, very aesthetic, high resolution, ultra-detailed, absurdres, scenery, photorealistic, {Prompt}, BREAK, depth of field, photorealistic details
Negative Prompts: modern, old, oldest, cartoon, graphic, text, painting, crayon, graphite, abstract, glitch, deformed, mutated, ugly, disfigured, long body, lowres, bad anatomy, bad hands, missing fingers, extra digits, fewer digits, very displeasing, (worst quality, bad quality:1.2), bad anatomy, sketch, jpeg artifacts, signature, watermark, username, simple background, conjoined, bad ai-generated
Description
Change the visuals and uprising the reality
NoobAI EPS v1.1 + Illustrious v2.0-stable DARE applied
FAQ
Comments (17)
Nice model. Very interesting results. Thanks.
"photography of a man" prompt, 4/4 female images (Illustrious v8.0)
Please use danbooru tags
For example, this case would be photo, 1boy
Samey face issue. Model makers should include more variety and avoid the pretty/insta filter bias
Great model! Love the quality
any idea on what settings to use to train a realistic person usable with this model ?
What's the difference between v9.0 and v9.0 DMD2?
Applying DMD2 LoRA
This allows it to create images faster by using LCM Sampler (Low steps, low CFG)
I've tested V9 DMD2 extensively. The images are incredibly beautiful, but for me, the best is Euler a/beta with 20 steps and CFG 3.0. Thank you for this, a very nice checkpoint. I love it
Absolutely cannot recreate preview results from V9, even when copying your entire generation parameters, mine always turn out with the same "bimbo face" (the typical middle-aged pornstar heavy make-up face)
Tried one, result not a million miles away
Depends entirely on what is he making those pics and on what you are. For example Forge vs ComfyUI will look a lot different, sometimes drastically.
@Mescalamba Thus the need to specify on which tool an image has been generated. Something that most showcased pics miss because of the author not taking time to have those infos displayed whatever the checkpoint.
yeah. A model might be able to produce nice results in theory, but if the showcase output requires the author's custom comfyUI (or whichever) workflow to generate anything close to that level of quality and prompt accuracy, then the model isn't nearly as good as advertised.
The quality of this model is still quite high, but it also misses the "happy balance" on just about every level. The detail and realism are good if you add a couple of effective realism loras, but it can't do ethnicity for shit. the character recognition, body shape control, and face shape pretty bad. The model struggles HARD with the same-face problem even with super heavy weighting on very well-known characters.
I'm out of the loop. What's DMD and what's the diff between the two v9s?
V9 is the direct updated version of v8 which got more variety with the face structures
DMD2 version is v9 with DMD2 LoRA applied
This version works on scheduler called LCM with only 8 steps, 1.5 CFG which normal version isn't able to
Although DMD2 works faster, some prompts that worked on normal one cannot work properly
So, in short: DMD2 version for faster creation, normal version for accurate creation
Absoltely goated model by a goated creator. This an nova asian are fantastic creations
Details
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.