This is an e621 model, use e621 tags. works well with furry artist tags (inside of fluffyrock) struggles with SD artists.
Redwater :
My original model and still my most loved, stick with this one if you want an overall model and use prompts to push it (say towards realism)
still does adult stuff the best and is better at emulating artist styles/most affected by artists tags.
Redwater703 :
A push towards realism, the idea behind this one was to have a model that defaults towards realism and then you can use artist tags to make interesting combos.
-
Suggestions for generation (all links are Civitai links):
A negative e621 inversion is very useful, I highly recommend boring_e621_fluffyrock_v4 as this model is a fluffyrock model at its core
If that one is causing problems then BWU from furtastic is also very good.
1024x768 (or vice versa) is the max I would go for txt2img generation, you're better off with limiting either dimension to 768 to prevent weirdness and then upscaling.
Credits:
WithoutOrdinary - For their fluffyrock model, this model uses fluffyrock-576-704-832-960-1088-lion-low-lr-e86-offset-noise-e71.
Freek22 - For their Newdawn model, this model uses newdawn_52aurora.
DominoPrincip - For their Opiate model, this model uses opiate_V30PrunedFp16.
A_B - For making stable diffusion fun and the advice on merging ^^ - A_B's models are here.
I recommend taking a look at the above models/creators and giving their stuff a try and a thumbs up.
Edits/updates:
Removed walls of text from description.
Removed advice on human in negative, I'm finding that it reduces the quality of the picture at least sometimes.
Description
FAQ
Comments (31)
Well, that's my second fav model around here for now. It really is great. Has interesting tricks up it's sleeve and was capable of doing things no other model ever did... yet. Why 4/5 then? Well, because the model has some serious relistic/semi-realistic influence to it, mandatory one at that. I haven't tried putting "realistic" tags into negative prompt yet, but I generaly prefer a model that doesn't require additional shamanic dancing with a drum around it. Yeah, I am a lazy b***h xD
Thanks for the review and awesome pics JxC. ^^
@Gungie thank you very much, Gungie! And thank you for the model again! Now I can free up my SD folder space a bit, because it really beats B95 ones... and I have several of them xD
PS I have more of 'em coming, just prefer to post them in sets/story-sets ;) the ones you saw are WIP. And yes, most of them will contain my human OC. Trying to recreate some of my old fics and RP sessions and "anthro on human" stuff is a personal drug of mine xD
Can this model generate human characters at all, or is it only animals? :)
It can. :)
I haven't tried myself but JxC is our resident expert and made plenty of awesome pics to demonstrate. ^^
@Gungie heh, you humble me. I just play around with characters and ideas in my head. My prompts aren't even that complex or thought out, but thank you nonetheless 😉
As I said before, yes, it can do humans. Although, it may still give you an anthro or nekomimi-type of character regardless of your prompt/negative prompt - but it happens with any furry-related model. Just play around with seeds and weights, CFG is also a great factor in this. For this model specifically if you really want to get humans, I wouldn't recommend very high CFG, 10 at maximum. If you like a certain image enough, it is worth it to tinker with it through inpainting or even outpainting. Oh, and sfw and nsfw are both an option.
Of course, this model has the data of mostly furry artists, but I also tried Frank Frazetta and Luis Royo on it and the results are pretty good.
Good luck.
@JxC Ah okay, that's a fairly long winded and verbose answer though. So human characters can't achieved by just typing 'human male/female' then? Perhaps merging with another model would be the best way to go about it. But yes, I'll play around and see what I can achieve. Cheers 👍
@SLAPY7 no, I didn't say that.
What I wanted to say is that you need to specify "human male/female" in the prompt, otherwise you can get anyone and anything.
If you really don't want any furry-related content, you should also specify so in the negative prompt - "furry, anthro, fur on human, human with animal ears", etc.
That's all.
i would like a feral animal/furry lora. you take comission ?
Sadly I've never done a lora and wouldn't know where to start. ^^
Honestly, I don't know why you even need one... I mean, ANOTHER one. I saw several of them on Civitai and even more off-site. Besides, this model and, for example, fluffymix_40 can do ferals by themselves, without any loras. It's just a matter of proper wording and try outs with different seeds. Believe me, I tested it... just can't post any results 😏
i ve been testing this one all day just discovered recently, yes is a very powerful model, easy to use, but be careful is a wild nsfw one, it can create anything you throw at it
This model is fantastic!
It has the ability to generate high quality and VERY sharp images. Amazing work!! (Redwater_703)
EDITED: This model is unique! It has a different noise pattern than I have seen. It sems have a "noise offset" pattern (I recommend to use "cfg rescale" in some cases)
😃
Thanks A_B ^^ truth is its an odd model. I'll have to try cfg rescale with it though I never thought to try. ^^
I have noticed that this model all but ignores OpenPose ControlNet functions. Any idea why or how to work arround it?
Oh I had no idea it did something like this, I'm guilty of only using openposedw to control hands. As for why, again not sure its a 3 model merge and doesn't do anything out of the ordinary.
I'm sorry Inusanfox I've no idea why or how to fix it but, If I learn anything I will post it here. ^^
I don't know if it's stated anywhere but do we need a VAE for this or is it already baked in? Awesome model btw, definitely one of my all time favorites for realistic looking anthros and such
hey ^^ You will want a VAE in general but any is fine, I believe without a VAE things will look washed out like most models.
There is no VAE baked in, I wanted folks to be able to choose freely and I often switch between an anime one or the original (the anime one is brighter... at least it feels so)
I hope this helps and thank you for the kind words! ^^
Thanks for your work! Is there a inpaint version of this model?
Maybe there is another model that can work as inpaint model with this one?
Hey Pereponok ^^ While I never made an inpaint model I did read that its fairly easy to make one :
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/How-to-make-your-own-Inpainting-model
The above details how to make one for any model.
I put one together for redwater using the above information and uploaded it here :
https://www.mediafire.com/folder/jsjmojfpjmud1/Redw-inpaint
I haven't done inpainting for quite a while and I've not tested the above model, if it works great if not write it off as redwater not having a inpaint model. ^^
Good luck! ^^
@Gungie Hope one day you launch a inpaint one. Loved your work here.
Hi @Gungie. Thank you for your reply. I know this method, but since the vpred models have their own specifics, I asked just in case ;)
And thank you for the uploaded model!
@pereponok268 no worries, I briefly played with it and it seems ok. so fingers crossed its good to go. Sadly I know of no other way of making one. ^^
its my favorite model with exellent result , will it be updated for more pose result ?
Hey Nicobay sorry about the late reply. ^^
Every time I try to update redwater I'm not happy with the results, therefore I kind of consider redwater to be done.
I'm very slowly working on new models though, so one day new ones might appear. ^^
I hadn't thought about poses so I'll keep that in mind, thank you. ^^
@Gungie aaaw , too bad , but thk you !
But, but, but soon SD3 cries
Can we expact any udpdates for this model? Or is it mostly done?
sd3.5 is highly trainable]e btw ;)
Lovely model but a shame we no longer get updates, there are a lot of tags that get ignored or misinterpreted sadly