Spaces:

stabilityai

/

stable-diffusion

like 10.3k

Running on CPU Upgrade

Hi @IamMrX . Just catching up on what has been posted within the last day. I am using the following text but also tweaking it on the fly depending on the topic. Hope it helps you.

Negative prompt text: out of frame, lowres, text, error, cropped, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, out of frame, 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, username, watermark, signature,

Also adding (((xxx))) brackets around text for emphasis helps, ... especially (((out of frame))) if you are getting partial heads.

@Crescent1967

Thanks :)))
btw are you getting better results with the negative prompts , i think the hands and fingers seem to be better with it

Yes, I agree... though I would prefer not to have to use them at all, there is some value.

In a perfect world you shouldn't need to mention things that a person would normally not expect to see... like deformities. I hope that things will get better so that we can use them to just exclude key things, like certain shapes, colors, or objects from a scene -- focus on offense not defense 😁

Yes, I agree... though I would prefer not to have to use them at all, there is some value.

In a perfect world you shouldn't need to mention things that a person would normally not expect to see... like deformities. I hope that things will get better so that we can use them to just exclude key things, like certain shapes, colors, or objects from a scene -- focus on offense not defense 😁

same here , only use them if the ai is not giving satisfactory results
besides this is just the beginning, things are only getting better and more advanced by the day :))

Excellent post @IamMrX
What about frames, there are times when AI will put a frame around an image.

For example: and it's always cut off like that.

Hi @AnySue . I have the same issue. Try adding "picture-frame" as a negative prompt. Just thought of it, but plan on trying it myself 😁.

For example: and it's always cut off like that.

i havent come across a frame yet , but as @Crescent1967 says, do try the negative prompt and see if it works, additionally you can try guiding the Ai with composition such as -- rule of thirds, golden ratio, perfect composition
also im guessing the ai is giving frame to a particular style you are putting or a specific artist style

Hi @AnySue . I have the same issue. Try adding "picture-frame" as a negative prompt. Just thought of it, but plan on trying it myself 😁.

That should work! I will for sure try it, thanks! 😃

@Crescent1967 , @IamMrX , have you tried Magic Diffusion? I like it so far.

@Crescent1967 , @IamMrX , have you tried Magic Diffusion? I like it so far.

nah not yet , is it free ? 😁😂😂

@Crescent1967 , @IamMrX , have you tried Magic Diffusion? I like it so far.

Nope. I haven't tried 'Magic Diffusion', but tried 'MagicPrompt Stable Diffusion' in the past, which is somewhat similar for generating prompts, without the art creation piece. Link: https://huggingface.co/spaces/Gustavosta/MagicPrompt-Stable-Diffusion

@AnySue , one thing that I do enjoy using on occasion is 'Clip Interrogator' (link: https://huggingface.co/spaces/pharma/CLIP-Interrogator). For example, generate an image yourself or take someone else's image and run it through clip interrogator and it will come up with some prompt text based on that image, use that prompt in Stable Diffusion to come up with a new set of 4 images,... save the best one and put that image into Clip Interrogator,... and so on and so on. It is rather interesting what prompts come up and where each will take you. Enjoy! : )

@Crescent1967 , @IamMrX , have you tried Magic Diffusion? I like it so far.

nah not yet , is it free ? 😁😂😂

It's free...😁😂😂

@AnySue , one thing that I do enjoy using on occasion is 'Clip Interrogator' (link: https://huggingface.co/spaces/pharma/CLIP-Interrogator). For example, generate an image yourself or take someone else's image and run it through clip interrogator and it will come up with some prompt text based on that image, use that prompt in Stable Diffusion to come up with a new set of 4 images,... save the best one and put that image into Clip Interrogator,... and so on and so on. It is rather interesting what prompts come up and where each will take you. Enjoy! : )

😃 That's what brought me to HF. I didn't like the ones on Google Colab. This site is so much nicer. 😊

Here is another fun option that I just stumbled upon - Stable Diffusion Prism. Link: https://huggingface.co/spaces/pharma/sd-prism

will give these a try

@IamMrX and @AnySue , here is another that I like playing around with with a bunch of different model options to choose from. https://huggingface.co/spaces/anzorq/finetuned_diffusion

@Crescent1967 I hadn't tried Stable Diffusion Prism before, I did just now though and I got 'Queue is full! Please try again.' I guess I'll try back, I'll add it to my likes so that I don't forget. 😄

I'm late to the party but --

Using 'bad composition' seems to fix a lot of issues.

I generally use these negatives:
bad anatomy, bad composition, ugly, abnormal, unrealistic, double, contorted, disfigured, malformed, amateur, extra, duplicate

Thank you @TheLonelyResearcher ... not too late :D

I'm late to the party but --

Using 'bad composition' seems to fix a lot of issues.

I generally use these negatives:
bad anatomy, bad composition, ugly, abnormal, unrealistic, double, contorted, disfigured, malformed, amateur, extra, duplicate

Thanks :))

give me a moment to find mine

(((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), fused fingers, messy drawing, broken legs censor, censored, censor_bar, multiple breasts, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2), black-white, bad anatomy, liquid body, liquidtongue, disfigured, malformed, mutated, anatomical nonsense, text font ui, error, malformed hands, long neck, blurred, lowers, low res, bad anatomy, bad proportions, bad shadow, uncoordinated body, unnatural body, fused breasts, bad breasts, huge breasts, poorly drawn breasts, extra breasts, liquid breasts, heavy breasts, missingbreasts, huge haunch, huge thighs, huge calf, bad hands, fused hand, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, fusedears, bad ears, poorly drawn ears, extra ears, liquid ears, heavy ears, missing ears, old photo, low res, black and white, black and white filter, colorless, (((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), fused fingers, messy drawing, broken legs censor, censored, censor_bar, multiple breasts, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2), black-white, bad anatomy, liquid body, liquid tongue, disfigured, malformed, mutated, anatomical nonsense, text font ui, error, malformed hands, long neck, blurred, lowers, low res, bad anatomy, bad proportions, bad shadow, uncoordinated body, unnatural body, fused breasts, bad breasts, huge breasts, poorly drawn breasts, extra breasts, liquid breasts, heavy breasts, missing breasts, huge haunch, huge thighs, huge calf, bad hands, fused hand, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, fused ears, bad ears, poorly drawn ears, extra ears, liquid ears, heavy ears, missing ears, old photo, low res, black and white, black and white filter, colorless, (((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), fused fingers, messy drawing, broken legs censor, censored, censor_bar, multiple breasts, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2), black-white, bad anatomy, liquid body, liquid tongue, disfigured, malformed, mutated, anatomical nonsense, text font ui, error, malformed hands, long neck, blurred, lowers, low res, bad anatomy, bad proportions, bad shadow, uncoordinated body, unnatural body, fused breasts, bad breasts, huge breasts, poorly drawn breasts, extra breasts, liquid breasts, heavy breasts, missing breasts, huge haunch, huge thighs, huge calf, bad hands, fused hand, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, fused ears, bad ears, poorly drawn ears, extra ears, liquid ears, heavy ears, missing ears, (((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), fused fingers, messy drawing, broken legs censor, censored, censor_bar, multiple breasts, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2), black-white, bad anatomy, liquid body, liquidtongue, disfigured, malformed, mutated, anatomical nonsense, text font ui, error, malformed hands, long neck, blurred, lowers, low res, bad anatomy, bad proportions, bad shadow, uncoordinated body, unnatural body, fused breasts, bad breasts, huge breasts, poorly drawn breasts, extra breasts, liquid breasts, heavy breasts, missingbreasts, huge haunch, huge thighs, huge calf, bad hands, fused hand, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, fusedears, bad ears, poorly drawn ears, extra ears, liquid ears, heavy ears, missing ears,

(((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), fused fingers, messy drawing, broken legs censor, censored, censor_bar, multiple breasts, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2), black-white, bad anatomy, liquid body, liquidtongue, disfigured, malformed, mutated, anatomical nonsense, text font ui, error, malformed hands, long neck, blurred, lowers, low res, bad anatomy, bad proportions, bad shadow, uncoordinated body, unnatural body, fused breasts, bad breasts, huge breasts, poorly drawn breasts, extra breasts, liquid breasts, heavy breasts, missingbreasts, huge haunch, huge thighs, huge calf, bad hands, fused hand, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, fusedears, bad ears, poorly drawn ears, extra ears, liquid ears, heavy ears, missing ears, old photo, low res, black and white, black and white filter, colorless, (((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), fused fingers, messy drawing, broken legs censor, censored, censor_bar, multiple breasts, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2), black-white, bad anatomy, liquid body, liquid tongue, disfigured, malformed, mutated, anatomical nonsense, text font ui, error, malformed hands, long neck, blurred, lowers, low res, bad anatomy, bad proportions, bad shadow, uncoordinated body, unnatural body, fused breasts, bad breasts, huge breasts, poorly drawn breasts, extra breasts, liquid breasts, heavy breasts, missing breasts, huge haunch, huge thighs, huge calf, bad hands, fused hand, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, fused ears, bad ears, poorly drawn ears, extra ears, liquid ears, heavy ears, missing ears, old photo, low res, black and white, black and white filter, colorless, (((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), fused fingers, messy drawing, broken legs censor, censored, censor_bar, multiple breasts, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2), black-white, bad anatomy, liquid body, liquid tongue, disfigured, malformed, mutated, anatomical nonsense, text font ui, error, malformed hands, long neck, blurred, lowers, low res, bad anatomy, bad proportions, bad shadow, uncoordinated body, unnatural body, fused breasts, bad breasts, huge breasts, poorly drawn breasts, extra breasts, liquid breasts, heavy breasts, missing breasts, huge haunch, huge thighs, huge calf, bad hands, fused hand, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, fused ears, bad ears, poorly drawn ears, extra ears, liquid ears, heavy ears, missing ears, (((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands), fused fingers, messy drawing, broken legs censor, censored, censor_bar, multiple breasts, (mutated hands and fingers:1.5), (long body :1.3), (mutation, poorly drawn :1.2), black-white, bad anatomy, liquid body, liquidtongue, disfigured, malformed, mutated, anatomical nonsense, text font ui, error, malformed hands, long neck, blurred, lowers, low res, bad anatomy, bad proportions, bad shadow, uncoordinated body, unnatural body, fused breasts, bad breasts, huge breasts, poorly drawn breasts, extra breasts, liquid breasts, heavy breasts, missingbreasts, huge haunch, huge thighs, huge calf, bad hands, fused hand, missing hand, disappearing arms, disappearing thigh, disappearing calf, disappearing legs, fusedears, bad ears, poorly drawn ears, extra ears, liquid ears, heavy ears, missing ears,

oh my ...😁✌

I call this type of prompting "beg crafting". You're essentially begging stable diffusion for what you want to see when in fact all you're really doing is introducing some random factors into how it denoises. I am firmly convinced that these types of prompts do basically nothing. The effects of prompt changes are statistical; therefore you have to generate many results from a prompt to be able to tell if it had a statistically significant effect. I think that people make changes to their prompt to beg stable diffusion for what they want ("award winning" being a particularly humorous example) and then after generating only a few results, if the randomized results happen to correlate with what they think they were accomplishing, they retain the prompt believing it was the cause. When in fact 99% of the time it was just the seed plus random effects from the prompt change.

I have a script that tries to fix hands in some generated art. It uses clipseg to create a mask around hands and then runs a loop generating new inpainted versions of the input image. I do not believe that I can influence how well it draws hands; I can simply have it generate a ton of variants and weed through them to find the best random result.

I added negative prompting influenced by the O.P.'s comment: "poorly drawn hands, poorly rendered hands, bad composition, mutated body parts, disfigured, bad anatomy, deformed body features, deformed hands".

This had basically zero effect on the quality of hand generation. Sure it made every seed produce a different result, but they were no better or worse than without that text included. I also tried the stable diffusion 1.5 model and the 1.5 inpainting model. The inpainting model was horrendous, mostly producing complete garbage, but once out of about 20 tries it produced a nearly perfect hand. That was unexpected, and I think just random chance. I think that the reason that the inpainting model did so poorly is because the original image was generated with a different checkpoint and the inpainting checkpoint wanted to see "very different things" than the original checkpoint would have seen.

I'll include 15 results from each run. First 15 is without the additional negative prompting. Next 15 is with the additional negative prompting. Final 15 is with the inpainting model.

If there's a way to put these images behind a spoiler tag so that they're only shown if someone wants to see them, please let me know how to do that. I am not familiar with the forum software being used on this site.

Original image:

No negative hand prompting:

Negative hand prompting:

Negative hand prompting, using inpainting model:

@bji , nice points there, i agree sometimes or most times the negative prompts don't really work , to be honest I don't even put them unless I need them to,
there is quite some experimenting to be done until we get some good results, unlike other models this one really needs lot of prompting and guides to get a decent result,
I haven't used the inpainting and outpainting features yet, would definitely like to try those out soon

@bji , nice points there, i agree sometimes or most times the negative prompts don't really work , to be honest I don't even put them unless I need them to,
there is quite some experimenting to be done until we get some good results, unlike other models this one really needs lot of prompting and guides to get a decent result,
I haven't used the inpainting and outpainting features yet, would definitely like to try those out soon

Negative prompts definitely have value; but I don't think that the model "understands" what people think it does. "Deformed hands" means nothing to the model. It just pushes things in a random direction.

@bji , nice points there, i agree sometimes or most times the negative prompts don't really work , to be honest I don't even put them unless I need them to,
there is quite some experimenting to be done until we get some good results, unlike other models this one really needs lot of prompting and guides to get a decent result,
I haven't used the inpainting and outpainting features yet, would definitely like to try those out soon

Negative prompts definitely have value; but I don't think that the model "understands" what people think it does. "Deformed hands" means nothing to the model. It just pushes things in a random direction.

Definitely

Stable was rendering images like this one using negative prompts:

So I tried the same text prompt with DALL-E 2 there isn't an option for negative prompts and I didn't use any in my text prompt and got this:

I usually like Stable over DALL-E 2 however, in this instance DALL-E 2 did a much better job.

Stable was rendering images like this one using negative prompts:

So I tried the same text prompt with DALL-E 2 there isn't an option for negative prompts and I didn't use any in my text prompt and got this:

I usually like Stable over DALL-E 2 however, in this instance DALL-E 2 did a much better job.

I feel sometimes the negative prompt in SD makes no improvement or sometimes even makes it bad
a big diff indeed , SD can produce good ones only when given lot of prompting lol , i hope in the future they upgrade their Data set or at least combine it with another one , that would be awesome

Stable was rendering images like this one using negative prompts:

So I tried the same text prompt with DALL-E 2 there isn't an option for negative prompts and I didn't use any in my text prompt and got this:

I usually like Stable over DALL-E 2 however, in this instance DALL-E 2 did a much better job.

Did you generate 100 images for each tool and compare the frequency of good versus bad results? Because if not, your results could just be random luck.

@bji I do not, that's a great point. I'm afraid I don't possess the patience required.

@bji was just wondering about the script you mentioned that uses clipseg to create a mask, is that something that's publicly available, I would love to try it.

even Adding following these words in Negative Prompt I often get double head, Double Body, Disfugred body,
instead, I get more often weird results.
in 50 attempts I get 2 or 3 comes proper result rest I get attached one.

Double head, Double figure, double body, Disfigured body, logo, watermark, watermark, text, title, signature, words, letters, characters, subtitle, cropped, zoomed, extra fingers, extra limbs, unnatural hands, extra legs, disfigured, disfigured fingers, disfigured hands, glasses, straws

Please help me to solve this issue🙏

@ameets21 ,

weird after we ve come a long way , SD still gives double heads and stuff like that , but then this is the base model without any filters, issue s will still pop up here and there , keep trying with and without, i remember most times negative made things worst except improving one or 2 issues at most

even Adding following these words in Negative Prompt I often get double head, Double Body, Disfugred body,
instead, I get more often weird results.
in 50 attempts I get 2 or 3 comes proper result rest I get attached one.

Double head, Double figure, double body, Disfigured body, logo, watermark, watermark, text, title, signature, words, letters, characters, subtitle, cropped, zoomed, extra fingers, extra limbs, unnatural hands, extra legs, disfigured, disfigured fingers, disfigured hands, glasses, straws

Please help me to solve this issue🙏

Stable Diffusion is trained at low resolution images (around 600 or 800 pixels). When you try to generate an image at 1200px, it kinda tries to fit 2 images inside it. Therefore, 2 heads.
Just keep the size (height and width) at about 600 or 800 pixels and I bet your aberrations will be solved! :)
If you want larger images, you can make some up-scaling technique.

Just keep the size (height and width) at about 600 or 800 pixels and I bet your aberrations will be solved! :)
If you want larger images, you can make some up-scaling technique.

That will help with double features, but has nothing to do with gnarled or malformed features, which are the vast majority of problems.

EDIT: Oh I see you are replying to the person who was having double-head issues; yes your advice is very sound in that case. I thought you were replying to the topic in general, sorry.

I will add woman and girl normally, but here it is: ugly, old, fat, slight, disgusting, Unflattering, Distorted, Poorly lit, Blurry, Grainy, Overexposed, Underexposed, Cluttered background, Distracting elements, Inconsistent theme, Unfocused, Harsh shadows, Glaring highlights, Mismatched colors, Unbalanced composition, Crooked, Tight framing, Excessive noise, Low contrast, Washed out, Oversaturated colors, Undersaturated colors, Awkward pose, Forced expression, Unnatural skin tones, Red-eye, Motion blur, Incorrect white balance, Uncomplementary background, Obstructed view, Uncoordinated color scheme, Dull lighting, Unappealing angle, Lack of depth, Disproportionate features, Unsympathetic expression, Incoherent theme, Overwhelmed by props, Lacking focus point, Disjointed composition, Inconsistent lighting, Unharmonious colors, Overprocessed, Underprocessed, Lacks emotion, Non-engaging, Stereotypical, Clichéd, Predictable, Uninspired setting, Misaligned elements, Disconnected from viewer, Unoriginal, Repetitive theme, Jarring contrasts, Uncoordinated elements, Stark lighting, Excessive editing, Unnatural enhancements, Overdone effects, Lack of authenticity, Artificial atmosphere, Forced theme, Unintentional subject, Unrelated elements, Incongruous style, Misplaced focus, Inappropriate mood, Overstated drama, Exaggerated features, Lack of subtlety, Inharmonious composition, Unnecessary distractions, Overcrowded, Simplistic, Unrefined technique, Lacking sophistication, Amateurish, Inelegant, Non-artistic, Lacks narrative, Unconvincing portrayal, Unpolished, Unprofessional, Inconsistent with theme, Muddled colors, Uncomplimentary lighting, Inadequate exposure, Imbalanced elements, Disproportionate composition, Overlooked details, Lacks depth of field, Unremarkable, Forgettable, Unimaginative, Monotonous, Tiresome, Uninspiring, Tedious, Lackluster, Unattractive, Aged, Overweight, Underweight, Repulsive, Disheveled, Scarred, Detestable, Unpleasant, Wrinkled, Emaciated, Unflattering, Misshapen, Gaunt, Sickly, Disharmonious, Distorted, Ragged, Disfigured, Haggard, Diseased, Sallow, Withered, Ghastly, Grotesque, Unsightly, Bedraggled, Weathered, Frail, Grizzled, Splotchy, Disordered, Unkempt, Blotched, Mismatched, Pallid, Lackluster, Faded, Jaded, Pockmarked, Creased, Weather-beaten, Worn, Tired, Bleak, Drab, Grim, Shabby, Raggedy, Rundown, Decrepit, Frazzled, Ruffled, Cluttered, Muddled, Unbalanced, Contorted, Lopsided, Crooked, Skewed, Mismatched, Uncoordinated, Inharmonious, Jarring, Clashing, Discordant, Asymmetrical, Uneven, Irregular, Unruly, Chaotic, Disheveled, Tousled, Messy, Disorderly, Tangled, Knotted, Snarled, Matted, Disarrayed, Scattered, Disorganized, Confused, Bewildered, Baffled, Perplexed, Flustered, Agitated, Disturbed, Troubled, Disconcerted, Unsettled, Discomposed, Dismayed, Alarmed, Perturbed, Anxious, Worried, Fretful, Edgy, Nervous, Tense, Strained, Stressed, Overwrought, Frantic, Hysterical, Panicked, Frenzied, Desperate, Fearful, Terrified, Horrified, Apprehensive, Distrustful, Suspicious, Paranoid, Cynical, Skeptical, Doubtful, Incredulous, Unbelieving, Nonbelieving, Disbelieving, Distrustful, Mistrustful, Untrusting, Guarded, Wary, Cautious, Prudent, Circumspect, Vigilant, Watchful, Alert, Attentive, Mindful, Observant, Scrutinizing, Analytical, Inquisitive, Curious, Questioning, Probing, Prying, Intrusive, Interfering, Meddling, Nosey, Snoopy, Intrusive, Overbearing, Domineering, Authoritarian, Tyrannical, Dictatorial, Autocratic, Oppressive, Heavy-handed, Overpowering, Coercive, Forceful, Compelling, Constrained, Restrained, Inhibited, Reserved, Withdrawn, Shy, Timid, Bashful, Reticent, Introverted, Solitary, Reclusive, Hermitic, Secluded, Isolated, Lonely, Lonesome, Forlorn, Abandoned, Deserted, Forsaken, Estranged, Alienated, Detached, Disconnected, Distant, Remote, Apart, Separated, Divided, Split, Disunited, Scattered, Fragmented, Fractured, Shattered, Broken, Cracked, Damaged, Flawed, Defective, Faulty, Imperfect, Deficient, Inadequate, Insufficient, Lacking, Wanting, Short, Missing, Absent, Omitted, Excluded, Left out, Overlooked, Neglected, Ignored, Disregarded, Overpassed, Bypassed, Skipped, Missed, Forgotten, Lost, Misplaced, Disappeared, Vanished, Gone, Erased, Wiped out, Obliterated, Destroyed, Demolished, Ruined, Ravaged, Devastated, Wrecked, Smashed, Shredded, Torn, Ripped, Cut, Slashed, Gashed, Pierced, Stabbed, Lanced, Sliced, Chopped, Severed, Cleft, Split, Divided,

I love how this space is still going on , cheers :))

Speaking to the model I have the most familiarity with (SD 1.5):

Most negative prompts do nothing -- they're just changing the noise generated from the seed.

However, SOME negative prompt words do have a STRONG effect, for example:

NSFW, naked, nude

However, these drastically limit the generated image's potential artistic style space. So... you may find that you're better off generating more images, and just throwing away the bare-breasted ones (if that's not what you're looking for).

I pretty much never use any any negative prompting with SD 1.5... I just generate more images -- usually 64 at a time once I'm generally happy with a prompt... that takes only 6 minutes on my 1070ti, and I can examine each while I generate the next batch of 64 (my 4070ti super arrives next week)... and pick out and save the 1-4 winners worthy of further consideration/time.

Sometimes I'll get an image that has awesome composition and body structure/detail, but the head is cropped off -- It may be worthy of saving, to later spend time to fix it with outpainting. These can potentially turn out truly amazing, but will require a lot of work. It's a little strange that these types (head crop-offs) have so much potential.

I will add woman and girl normally, but here it is: ugly, old, fat, slight, disgusting, Unflattering, Distorted, Poorly lit, Blurry, Grainy, Overexposed, Underexposed, Cluttered background, Distracting elements, Inconsistent theme, Unfocused, Harsh shadows, Glaring highlights, Mismatched colors, Unbalanced composition, Crooked, Tight framing, Excessive noise, Low contrast, Washed out, Oversaturated colors, Undersaturated colors, ...

Complete waste of time.

I love how this space is still going on , cheers :))

It's been a long time!! I love that it's still going on too. We've known each other for a while, lol. See you on NC, FB, and Messenger sometime soon. xo

I love how this space is still going on , cheers :))

It's been a long time!! I love that it's still going on too. We've known each other for a while, lol. See you on NC, FB, and Messenger sometime soon. xo

we ve had some great times here , starting our Ai journey