Check my exclusive models on Mage: ParagonXL / NovaXL / NovaXL Lightning / NovaXL V2 / NovaXL Pony / NovaXL Pony Lightning / RealDreamXL / RealDreamXL Lightning
Recommendations for using the Hyper model:
Sampler = DPM SDE++ Karras or another / 4-6+ steps
CFG Scale = 1.5-2.0 (the lower the value, the more mutations, but the less contrast)
I also recommend using ADetailer for generation (some examples were generated with ADetailer, this will be noted in the image comments).
This model is available on Mage.Space (main sponsor).
You can also support me directly on Boosty.
Realistic Vision V6.0 (B2 - Full Re-train) Status (Updated: Apr. 4, 2024):
- Training Images: +3400 (B1: 3000)
- Training Steps: +724k (B1: 664k)
- Approximate percentage of completion: ~30%All models, including Realistic Vision (VAE / noVAE) are also on Hugging Face
ᅠ
Please read this! How to remove strong contrast.
To make the image less contrasty you can use LoRA [Detail Tweaker LoRA] in a negative value.
ᅠ
Orange Color = Optional
ᅠ
I use this template to get good generation results:
ᅠ
Prompt:
RAW photo, subject, 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3
ᅠ
Negative Prompt:
(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime), text, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, UnrealisticDream
(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime, mutated hands and fingers:1.4), (deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, disconnected limbs, mutation, mutated, ugly, disgusting, amputation, UnrealisticDream
ᅠ
Euler A or DPM++ SDE Karras
CFG Scale 3,5 - 7
Hires. fix with 4x-UltraSharp upscaler
Denoising strength 0.25-0.45
Upscale by 1.1-2.0
Clip Skip 1-2
ENSD 31337
ᅠ
Thanks to the creators of these models for their work. Without them it would not have been possible to create this model.
HassanBlend 1.5.1.2 by sdhassan
Uber Realistic Porn Merge (URPM) by saftle
Protogen x3.4 (Photorealism) + Protogen x5.3 (Photorealism) by darkstorm2150
Art & Eros (aEros) + RealEldenApocalypse by aine_captain
Dreamlike Photoreal 2.0 by sviasem
HASDX by bestjammer
Analog Diffusion by wavymulder
Life Like Diffusion by lutherjonna409
Analog Madness by CornmeisterNL
ICBINP - "I Can't Believe It's Not Photography" by residentchiefnz
Description
This model does not include VAE.
This version of the model focuses on photorealism and may be worse than version 2.0 in some aspects.
FAQ
Comments (52)
Why does EVERYONE have cleft chin? :( Despite of adding it to the negative prompts.
(Experts say that there are only 6% of people with cleft chin)
I blame the cloud people. These ethereal beings possess an uncanny ability to subtly inject their inclinations into the greatest models.
More info from: nevervaccinate.com
Probably doesn't understand what cleft chin is....why negative doesn't work
I would download it, especially for that cleft chin, ❤️
1.3 is better than v3 . this is too platic...y
@SmartNeto When 1.3 came it was my main model. it was so realistic. Now i can find a dozen of models look like v3. they all look the same. v1.3 is the best
Which VAE should I use with version 1.3? Any help?
@lomewa5023 Hi! You can go to the V3.0 noVAE model and download the VAE I have uploaded.
If I'm not mistaken, version 1.3 already comes with included VAE
@SG_161222 Thanks king
Do I need to disable existing VAE when this includes one?
Hi. That's not necessary.
@SG_161222 Thanks, I was disabling the standard 84000 VAE when switching models that had an embedded one because I was not sure... The more you know :)
Many thanks. I love your work and your model is my first one
Thank you for that, I really appreciate it!
Hello @SG_161222
on behalf of civit team, can you get in contact on the admins of Civit discord as soon as possible?
Mediterranean style,interior design
My realism training tutorial based on realistic vision model : https://youtu.be/g0wXIcRhkJk
What are the differences between the models on Civitai and Hugging Face? I don't quite understand. Thank you.
Hi! There's no difference.
@SG_161222 it appears there is a discrepancy with the licensing. We're interested in allowing our users to generate images using your models.
While HuggingFace's license allows generation services, this appears to be different on the Civitai page.
Moreover, there are some Realistic Vision models uploaded on Replicate.
Could you clarify your position on this?
Thanks for this; I appreciate your work. For some reason, female faces always look blurry, disfigured, or deformed. Is this normal when prompting characters in full body views or is there some setting or addition I'm overlooking?
Hi! You can use the After Detailer extension to enhance faces.
@SG_161222 Thanks for your response, and I'll try the extension today. So I take it it's normal for faces to be this way in Stable Diffusion? I just worried I might have downloaded your model incorrectly or something. :) Does this also mean I should disable the Face Restoration setting in Stable Diffusion or just leave it?
@MultifacetedGentleman If you are going to use the After Detailer extension, it is better to disable the Face Restoration option (you can also use After Detailer and Face Restoration at the same time).
The quality of the faces also depends on the generation resolution. You can also increase the resolution (I usually use 640x960). Hires. fix can also help in this case.
@SG_161222 Alright, I'll see how it goes. Thanks for getting back to me.
I cant convert inpainting to diffusers help plz
Hi! As far as I know the inpainting model cannot be converted to diffusers (correct me if I'm wrong). If you want to train an inpainting model, you must first train the standard model, then convert the standard model to an inpainting model using model merging.
I used Automatic1111. Use the merge with add difference.
Model A = SD 1.5-inpainting
Model B = Standard model
Model C = SD 1.5 pruned (7GB)
Multiplier = 1
I'm afraid I must agree with some of the negative comments here. As much as I appreciate the work of the model creator, version 3.0 doesn't provide satisfactory results compared to previous versions.
It's like taking a picture and adding lots of artificial noise and detail. The image looks completely artificial even with LoRAs such as <lora:more_details:-1> or playing with different analog camera models. But hey, maybe this is what you're looking for; what do I know?
See my comparison tests for reference: Image post by draw a line (civitai.com)
@SG_161222 Thanks so much for your answer, and good luck with SDXL!
@drawaline Thanks :)
The model is already uploaded on Hugging Face. If you would like to try this model, I would be interested to know what you think of it.
@SG_161222 playing with it today... wow, the results are amazing. I'll post some pictures this week, but this is by far the best realistic model I've used on my slow M1. I need to use <lyco:add_detail:-0.4> to get what I want, but the images are absolutely stunning with remarkably simple prompts. Thanks for your work on this! I can't wait for our SDXL release!
Do you have any Loras baked into this checkpoint, such as Lowra? All the images look way too dark.
can you like
I might be wrong but when using unrealisticdream in the negative prompt, shouldn't we been using baddream too? if I remember correctly on the embedding page it says that bad dream is standalone but unrealisticdream needs to go in pair with baddream. does it work anyway?
Hi all! A lot of people don't like version 3.0, so now I'm working on a new version (some images) without the heavy contrast and shadows. I think this will be the final version and I'll switch to SDXL. If you want to test the new update you can write me on Reddit. Thanks!
This is a lot better imo.
Hi all again! I think 4.0 is ready. First the model will be uploaded on Hugging Face in the version without the baked VAE, but the VAE will also be uploaded with the model. After a little while I will upload the model on CivitAI.
thank you! can't wait to give it a try!
Already downloading it!!
wonderful!
The best model till now! thank you
Hi i have noob question. Whats is diference in VAE and noVAE?
@Codexer Hi. The VAE file is already baked into the VAE version. In the noVAE version the VAE file is not baked and it must be downloaded separately and used with the model. That's in plain words.
I think something is wrong with the uploads of the unpruned safetensors files - 3.0 and 4.0 have identical SHA256 hashes (checked on both CivitAI and Hugging Face). This isn't the case for the pruned safetensors or unpruned ckpt.
@Civet_AI Thank you! Apparently I wasn't very attentive when I converted the models, I'll fix it now.

