This embedding will tell you what is REALLY DISGUSTING🤢🤮
So please put it in negative prompt😜
⚠This model is not trained for SDXL and may bring undesired results when used in SDXL.
If you use SDXL, recommended this 👇
another deep-negative:
pony version: https://civarchive.com/models/831971
SDXL version: https://civarchive.com/models/407448
TOP Q&A
how to use TI model?
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Textual-Inversion
what is negative prompt?
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Negative-prompt
[Special Reminder] If your webui reports the following errors:
- CUDA: CUDA error: device-side assert triggered
- Assertion -sizes[i] <= index && index < sizes[i] && "index out of bounds" failed
- XXX object has no attribute 'text_cond'
Please try using a model version other than 75T.
> The reason is that many scripts do not handle overly long negative prompt words (greater than 75 tokens) properly, so choosing a smaller token version can improve this situation.
[Update:230120] What does it do?
These embedding learn what disgusting compositions and color patterns are, including faulty human anatomy, offensive color schemes, upside-down spatial structures, and more. Placing it in the negative can go a long way to avoiding these things.
-
What is 2T 4T 16T 32T?
Number of vectors per token
[Update:230120] What is 64T 75T？
64T: Train over 30,000 steps on mixed datasets.
75T: embedding limit maximum size, training 10,000 steps on a special dataset (generated by many different sd models and special reverse processing)
Which one should choose?
75T: The most ”easy to use“ embedding, which is trained from its accurate dataset created in a special way with almost no side effects. And it contains enough information to cover various usage scenarios. But for some "good-trained-model" may hard to effect
and, change about may be subtle and not drastic enough.
64T: It works for all models, but has side effect. so, some tuning is required to find the best weight. recommend: [( NG_DeepNegative_V1_64T :0.9) :0.1]
32T: Useful, but too more
16T: Reduces the chance of drawing bad anatomy, but may draw ugly faces. Suitable for raising architecture level.
4T: Reduces the chance of drawing bad anatomy, but has a little effect on light and shadow
2T: ”easy to use“ like T75, but just a little effect
Suggestion
Because this embedding is learning how to create disgusting concepts, it cannot improve the picture quality accurately, so it is best used with (worst quality, low quality, logo, text, watermark, username) these negative prompts.
Of course, it is completely fine to use with other similar negative embeddings.
More examples and tests
draw building: https://imgur.com/5aX9yrP
hand fix: https://imgur.com/rDlsrgS
portrait (with PureErosFace): https://imgur.com/1Lqq595 https://imgur.com/V5kXBXz
fusion body fix:
How is it work?
I tried to make SD learn what is really disgusting with deepdream algorithm, the dataset is imagenet-mini (1000 images chosen randomly from the dataset again)
deepdream is REALLLLLLLLLLLLLLLLLLLLLY disgusting 🤮 and process of training this model really made me experience physical discomfort 😂
Backup
Description
put it in negative prompts
FAQ
Comments (15)
Any chance of making one for 2.1? 2.x-based models are the ones which suffer the most from distortions and which necessitate of negative prompt engineering the most.
will but not now~👌
I haven't found the right way to training it. truth be told,current training is actually aimless... (Im surprised it can actually work🤣)
v1 version is just an experimental output, now need more test more reseach
Testing V1 2T seems to cause small issues with eyes, maybe you could blur or smudge the eyes so it isn't creating such a strong negative towards normalish results.
and I thought that yesterday my models had miracles with their eyes)) All oblique, crooked, a look only in the vicinity is possible)) And I put this negative promt and forgot))
@minimalist what sd-model you using? i am testing it
@FapMagi Elldreth's Lucid Mix
I don't understand the weights combinations in your sample pictures.
Are you ask to the meaning of different coordinates? I did omit to write...
it is prompt-editing and attention weight
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Features#attentionemphasis
TL;DR: top-left is weakest effect, bottom-right is highest
If you're talking about "[( NG_DeepNegative_V1_64T :0.9) :0.1]", the square brackets signify that this part of the prompt is only added after 10% of the steps are done, in other words it has no effect for the first 2 steps if you have 24 steps or so. If you have 2 ":" in the brackets, then it switches from the first to the second part at the step indicated at the end. It is kind of confusing since it looks exactly like attention weights.
@wewewew Thanks for the response, would you mind explaining hot to achieve such matrix ? I guess one would use the S/R function in x/y/z plot script, but I can't figure an easy way how to so.
@IDEIYIONI I don't know what you mean, this isn't related to prompt matrix or scripts. It's a default feature that triggers normally. If you set the steps at 20 for example, a prompt is sent to the AI each of those 20 times, and say you put "picture of a [car:horse:0.2] in a field" as your prompt, the first 4 steps the AI will receive "picture of a car in a field", and then car will be switched to horse for the rest. You can enable the option to display in-progress images every step, and you might see a vague car switch into a horse, or you might not because it's pretty hard for the AI to modify the content that much unless you set it to change at step 2 or 3. Similarly you could put [car|horse] which switches it every step.
@wewewew i was just talking about this same confusion with a friend two days ago because he was using [prompt:n] as demphasis weights and almost convinced me it worked the same before going over docs
Trust me! Using NG_DeepNegative_V1_4T as the only (positive) prompt returns cute critters with aEros model!
WTF!!!🐵
I thought you were joking until I actually tried it just now. While the output is still disgusting, it's actually a whole creature😨
it Like the aliens I've seen in my dreams! (this is a joke)
aEros model is very very different from other
Post one :D