Real Dream Website (Faster download)
Available at TensorArt: https://tensor.art/models/899250342552672189/
Comfyui Workflows:
Anima text2image with Loras Workflow
Anima img2img (denoise) with Loras Workflow
Klein 9B Edit with Loras Workflow
Klein 9B img2img (denoise) with Loras Workflow
Z-image-turbo img2img workflow
Z-image text2img with loras workflow
Z-image De-Turbo with loras workflow
This model is capable of generating very realistic images that can confuse people's critical sense, please do not use it irresponsibly, harming ordinary people or public figures. Also avoid its use in electoral campaigns, promoting or damaging the image of candidates. Although the field of artificial intelligence is poorly regulated, its improper use can cause legal penalties even in countries with little regulation in the area. Be very careful to not commit crimes in your country and not to promote injustices against innocent people. I do not authorize the use of this model for Lora training using the face of real people without authorization from the biological owner. Do not create images that can embarrass and damage the public image of people. This can cause serious psychological harm, including deep depression. I do not authorize the use of images generated by this model for scam applications and the dissemination of false information.
During the fine-tuning of AI models, I only use synthetic images that do not infringe any copyright, personal image rights, or laws. The models are merges of open models from different sources and may contain diluted data remnants in the neural network. I am not responsible for possible data remnants since they are diluted, and it is not humanly possible to track the existence and origin of this diluted data.
I do not allow the upload of any of my models on Tensor Art. I only do not allow it there because the policy of duplicate models on Tensor Art prevents me from uploading again if someone has already done it.
I have enabled to automatically hide all NSFW posts for several reasons that are not worth listing. I hope you have patience and understand.
Description
FAQ
Comments (29)
Hey,thanks for this! I will say to anyone commenting, this model NEEDS to be on the settings of the example pics,pretty much. Do not stray far from that,and you should get nice pics. Also,im not an expert and could be wrong in some ways.
Hey Sinatra, you are merging master. This is my thought.
Buen modelo, no publico más post, ya no te molesto más.
Good model, I won't publish any more posts, I won't bother you anymore.
Por cierto, los nombres son inventados.
By the way, the names are made up.
amazing realistic photo can be created
one of the most realistic models for SD1.5 available on Civitai in January 2023.
You mean January 2024?
yes! thank you
You make bold claims in your description - yet all your example images are simple closeup face portraits that needed a whole essay for the prompt with negatives.
I did a lot of testing with simple prompts that tend to generate images that stable diffusion has a lot of difficulty generating. The model has performed better than most photorealistic models, I can start putting images here with simpler prompts. I've kind of gotten used to using complex prompts but there's not much need for it.
Is a single paragraph of text what passes as an essay these days? That would explain a LOT about the education system if so xD
I made the bold claim that after intensive testing this model may well be the best of all time in an article I published on the site, and I have 5 versions of it that all hang in with any of the greats. Did you make any gens to verify said bold claims do not add up?
@sinatra thankyou ! I think many people would be very interested in seeing that :)
@Xantim Nope. I'm pretty sure it's called a "metaphor"
@MysteryWrecked Firstly, I heavily disagree with your article but thanks for making it, it has lots of recommendations. Secondly, I haven't even downloaded the model, I'm simply commenting on the example images in relation to the big bold letters in the description. Thanks.
@Tigra I think 'gross exaggeration' is what you're looking for, but it's fine haha ;)
I posted a couple gens under v10, I'm not sure if they're showing up, but they are in my profile. They are rendered at 800x1200 with 180 steps, and have photo, realism, and detail loras applied, as well as various embeddings for the same purpose, and hi-res fix with face fix and upscaling as well, but no other model really does better than this in 1.5 regardless. I make 512 gens at 24 steps with no hi-res just to test my controlnet poses, and often they are so good that I will hi-res fix and upscale them for my collection. I really like this model.
Edit: All my images got taken down by the site, no reason, not reposting.
My take on this model is it's not in any way noticeably better than other 1.5 models simply because there's nowhere left to go with the extremely low quality fundamentals of 1.5. I think you would have to train a model where there were MORE new inputs that images that existed in the first place to truly create a groundbreaking model, at this point. And your inputs would have to be very very carefully selected, probably mass-cleaned of JPEG artifacting with a 1x ESRGAN model first too.
I'll help you with the Buzz if for no other reason than to oppose the AI snobs that shoot everyone down.
In return for the people who donated Buzz, I have already created a new Lora, with the focus on generating hands. https://civitai.com/models/257185
I'm studying the feasibility of merging with Real Dream.
Good results but very difficult to get non closeup portraits.
which prompt did you use? try using "full body shot"
@sinatra Yeah that can work but anything in between full body and portrait is really difficult. The default image 99 times out of 100 is a closeup portrait.
@someguy1927534 prompt "full-view" or "in-full-view" and you get what you are looking for. and those dashes between those words are VERY important! don't forget them. they work as weights.
appreciate your work
great real images
Tried it, it's ok. Not in any way noticeably better than other photoreal 1.5 models though. Like I'm largely convinced that there's nowhere left to go with 1.5 other than training a model where there are MORE new input images than images that existed in it originally, and where all of those new ones have been indivdually inspected for quality and probably mass-cleaned of compression artifacting with a 1x ESRGAN model first.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.