[Stable Diffusion] Prompt Sharing and Learning Thread

[Mr-Fox]

Thank you! How long does it take for you to bring a still image to a little "animated" gif? Can you post an example of one of your results?

I can't remember tbh, it was a long time since I experimented with it.

This was only a test I did to see how it would work with faceswap:

Source (right click and set to loop)	The Result.. (right click and set to loop)
View attachment ezgif.com-video-to-gif [MConverter.eu].webm	View attachment [MrFoxxx] Sukie Deep [MConverter.eu].webm

I made this by using the batch function in img2img. First open the source in photoshop, crop it to the right resolution and adjust the length of the video. Then export it as video frames (images). Put these in an input folder. Now you can use these video frames as input or source images in img2img and do what ever changes you wish, including using controlnet. Save the output images in an output folder. Now you need to put it back together into a video again. I used Flowframes. It can double the fps if you wish.
Tutorial:

You must be registered to see the links

Source (right click and set to loop)	The Result.. (right click and set to loop)
View attachment Alyx Star Big Tits Cut.webm	View attachment output-2x-RIFE-RIFE4.0-40fps.webm

Keep in mind that converting the files to webm is decreasing the quality.

 

[Fuchsschweif]

I made this by using the batch function in img2img. First open the source in photoshop, crop it to the right resolution and adjust the length of the video. Then export it as video frames (images). Put these in an input folder. Now you can use these video frames as input or source images in img2img and do what ever changes you wish, including using controlnet. Save the output images in an output folder. Now you need to put it back together into a video again. I used Flowframes. It can double the fps if you wish.
Tutorial:

You must be registered to see the links

I was mainly interested in how long it took you to do these - if I have to wait 2-3 minutes that would be fine, if it takes 10+ min then I would probably wait until I can upgrade to a better GPU

Other than that, I am still open for recommendations regarding websites offering such a service.. even though I couldn't find one, it's hard to believe for me that there is none out there.

 

[EvylEve]

Hey guys! I guess with my GTX 1070 gpu I can't even dream of generating image to video with SD (let me know if I am wrong).
So I am currently searching for a website where I can easily load in some NSFW images and just get them animated (just a bit of movement). Any idea where I can do that? I only find NSFW ai generators that offer text to video, but not image to video. And the big ones of course don't allow NSFW stuff.

Performance and VRAM will be the main bottlenecks. And probably it's better to stick to SD 1.5 (work 512x512 then upscale). If you have a fast CPU, it may be not so bad, but still slow.

Said this, you can probably get a Google Colab account for a few bucks / month and rely on their processing power, but I don't really know if they have some weird limit against NSFW content.

 

[Mr-Fox]

I was mainly interested in how long it took you to do these - if I have to wait 2-3 minutes that would be fine, if it takes 10+ min then I would probably wait until I can upgrade to a better GPU

Other than that, I am still open for recommendations regarding websites offering such a service.. even though I couldn't find one, it's hard to believe for me that there is none out there.

If 10 + minutes is too much for you then forget it...

 

[Mr-Fox]

Performance and VRAM will be the main bottlenecks. And probably it's better to stick to SD 1.5 (work 512x512 then upscale). If you have a fast CPU, it may be not so bad, but still slow.

Said this, you can probably get a Google Colab account for a few bucks / month and rely on their processing power, but I don't really know if they have some weird limit against NSFW content.

Hogwash.. I have a GTX1070 and I'm not limited to 512x512 images. You need to keep it under 1024 in most cases anyway then it's either hiresfix or upscaling in img2img that is next and I can get an image of 8k if I really want.
Of course going that big will take some time and is not necessary nor practical.
But to your point, yes Google colab or runpod is an option for those that want to do something demanding like making a video or training a model but lack the hardware for it.

 

[felldude]

The biggest limitation on training is the amount of VRAM.
If it won't fit it wont' fit.

A GTX 1070 should run around 90% of IT's per second as a RTX 3050 and
In theory a RTX 3070 would be around 3x faster.

However neither of those cards can do full dreambooth training on XL as that takes around 24GB of VRAM

The RTX 3070 24GB card would be my "low end" choice at about 1500$

Then the Quadro RTX 5000 at around 4000$
If you don't mind PNY

Then you have the tesla cards at around 8000$
Ironicaly Dell is the prefered vender and one of the only ones not selling fake cards.

 

[deadshots2842]

I need help can some one tell me how to write prompt to get this girl (superwomen from DC).

 

[felldude]

I need help can some one tell me how to write prompt to get this girl (superwomen from DC).

If your using Pony it would be danbooru tags.

But most other models you could use generic tagging such as:

1girl, solo, long hair, breasts, large breasts, black hair, gloves, ponytail, shiny, dark skin, dark-skinned female, lips, bodysuit, makeup, lipstick, skin tight, shiny clothes, latex, latex gloves

 

[devilkkw]

I have made a1111 like workflow for cui. This is for sd1.*.
I've made it for who want to try and understanding CUI.
Actually is simple but almost complete for text2image generation.
It have wildcard support and Lora loader.
It also use ToMe patch.
Tell me if you are interested and i do update by adding other feautures like upscaler, img2img, impaint.
Install

You must be registered to see the links

in your CUI and when load click on button "Manager" then "install missing node.

 

[Jimwalrus]

I need help can some one tell me how to write prompt to get this girl (superwomen from DC).

In addition to prompts, there are two LoRAs on Civitai for her (one is SD1.5 & one PonyXL).
I've tried the 1.5 one and it works well.

 

[deadshots2842]

If your using Pony it would be danbooru tags.

But most other models you could use generic tagging such as:

1girl, solo, long hair, breasts, large breasts, black hair, gloves, ponytail, shiny, dark skin, dark-skinned female, lips, bodysuit, makeup, lipstick, skin tight, shiny clothes, latex, latex gloves

Danbooru tags what's that can you teach me or point me to youtube videos? And yes I'm using pony xl with lora

 

[deadshots2842]

I have made a1111 like workflow for cui. This is for sd1.*.
I've made it for who want to try and understanding CUI.
Actually is simple but almost complete for text2image generation.
It have wildcard support and Lora loader.
It also use ToMe patch.
Tell me if you are interested and i do update by adding other feautures like upscaler, img2img, impaint.
Install

You must be registered to see the links

in your CUI and when load click on button "Manager" then "install missing node. View attachment 3590766

I used to have comfy then I deleted it and installed a1111 lol

 

[Fuchsschweif]

If 10 + minutes is too much for you then forget it...

I can't wait until we can create entire NSFW 3D animations that look perfect just by inputting prompts and generate a 30 sec clip in the blink of an eye..

 

[Fuchsschweif]

No idea, don't use it. It's like PC Master Race (auto1111 with 4090) vs console peasants (comfy) I don't hang out with the great unwashed!

It's the other way round, auto1111 equals consoles. Comfy has a better performance and offers also way more customization. Even the Stable Diffusion developers at OpenAI work with ComfyUI.

 

[felldude]

Danbooru tags what's that can you teach me or point me to youtube videos? And yes I'm using pony xl with lora

I have only ever trained one model on pony and I don't use it that often.

I would look through civitai for prompts as knowing an artist name is key to triggering the training data, along with the image scoring and general prompts

It's the other way round, auto1111 equals consoles. Comfy has a better performance and offers also way more customization. Even the Stable Diffusion developers at OpenAI work with ComfyUI.

I'd have to assume that your replying to a joke comment as the 40 series was marketed toward gamers with "special cores" the actual gpu cores and stats are not much different from the 30 series (Baring DLSS and DLAA marketed time improvements).

With both Comfy and Auto1111 being gooeys (GUI) the "true" PC master race does command line generation from hugging face diffusers lol

I ended up using comfyUI but my understanding is huggingface adopted Gradio as the backend for building GUI's so Auto1111 gets all the new stuff a lot easier then Comfy

 

[devilkkw]

I used to have comfy then I deleted it and installed a1111 lol

Personal choice, but why not keep both? i use a1111, forge and CUI. You can set it to use same folder for models,lora,embedding,etc.

 

[deadshots2842]

Personal choice, but why not keep both? i use a1111, forge and CUI. You can set it to use same folder for models,lora,embedding,etc.

Saving space. Got 1tb gen 3 mvme but with lot of games. I'll Add more storage later.

 

[deadshots2842]

I have only ever trained one model on pony and I don't use it that often.

I would look through civitai for prompts as knowing an artist name is key to triggering the training data, along with the image scoring and general prompts



I'd have to assume that your replying to a joke comment as the 40 series was marketed toward gamers with "special cores" the actual gpu cores and stats are not much different from the 30 series (Baring DLSS and DLAA marketed time improvements).

With both Comfy and Auto1111 being gooeys (GUI) the "true" PC master race does command line generation from hugging face diffusers lol

I ended up using comfyUI but my understanding is huggingface adopted Gradio as the backend for building GUI's so Auto1111 gets all the new stuff a lot easier then Comfy

Am I supposed to train them or I saw a video where he got the prompt from photo. Can I just use those prompt and get my result or do I have to train??

 

[felldude]

Am I supposed to train them or I saw a video where he got the prompt from photo. Can I just use those prompt and get my result or do I have to train??

You can get the prompts from the images, you don't have to train anything. Pony is already trained to certain artist.

An example from civitai

score_9, score_8_up, score_7_up, equestria_girls

Replace "equestria_girls" with any artist or category on rule34 and you have a start

 

[Sharinel]

Am I supposed to train them or I saw a video where he got the prompt from photo. Can I just use those prompt and get my result or do I have to train??

Also in addition to what felldude said, have a look here

You must be registered to see the links

You must be registered to see the links

Even if you don't use wildcards, you can use the .txt files as a database to see what charcters are incorporated into Pony without having to download loras