Disclaimer, this is not my script, I did not make it and I can't take credit for it whatsoever (if you recognise the script and it's owner, please let me know so I can contact them and ask them for permission, if you recognise this as your own script and you would like it removed, please let me know!)
The initial script was designed for making a deepthroat animation, and admittedly I could never get it to work, but it piqued my curiosity so I've tampered with it several times, this being one of the better iterations! This doesn't do anything the original script it will allow, so once again, the original author deserves all credit.
For anyone who knows how to edit the script, you'll be able to see what it does. This version has 18 frames, ranging from "topless, (small breasts:1.2), nipples" > "topless, (huge breasts:1.4), nipples) and exports them into a gif afterwards. I couldn't work out how to upload the file without choosing a .zip file, but just extract it into the 'Scripts' folder and it should show up where you'd choose the X/Y prompt option.
Advanced tips:
1: You should try to control the image as much as possible, making sure to pose your subject, their hands, the background as much as possible so as much will stay the same as possible.
2: Img2Img frames. If the gif turned out alright, save for one or two frames where it's a little too different, I've had decent luck using Img2Img with that frame, until it looks like it'll match with the rest. Then just use something like https://ezgif.com/maker to make it manually!
3: It prefers drawn models more than realistic!
Description
This doesn't quite work out as smoothly as the breast size increase does, but it does what I need it to. Even more convinced that it'll make a better timelapse vs transformation.
Scales from "(plump:1.0)" > "(abs, obliques, toned:1.4), (biceps:0.9)"
Gifs were manually editied and turned into a gif with https://ezgif.com/. You can see the difference crossfading has on the two images as well.
FAQ
Comments (71)
Hi, I guess you are talking about the script mizumizuni-concept-v1.5, unfortunately its account is deleted
@drtenta That's the one! As I said, I'm really not trying to claim credit, and if he'd rather it disappear I'll happily delete it, I just wanted to share what I managed to do with all his hard work was all.
@BDinc no worries on my side, I'm happy to see his work continue :D
@BDinc That was an alt account of mine, that the mods threw a fit over. You're fine to use or share any of my stuff. I don't care.
I didn't even really write the whole script; I got gpt-4 to do most of the work. :)
Also, love the idea! I'm surprised the mods haven't shut it down. They, uh, they kinda have it out for small breasts.
@pythagorveil Oh hey! Sorry to hear that, but thank you for the kind words! I tried to do a pregnancy timelapse, but that did not work very well whatsoever. I think it's cool, it certainly does some fun things!
@pythagorveil Will you republish your work on another platform?
@drtenta My stuff's all up on huggingface, I still have to upload my scripts there. I've got a few new loras to share; I haven't decided yet if I'm going to add them to huggingface, or post them under a new name here.
As for other platforms, there isn't really anything up yet. But when there is, I'll probably post on whatever comes up. I mean, that's assuming loras aren't obsolete by then :)
@pythagorveil Let us know what to search for when they're uploaded!
V2 will attempt a more gradual animation, with added emotions added to it. I also plan on doing a pregnancy timelapse if everything goes to plan!
in witch folder i have to put the script?
@Jaulustus in the main SD folder should be a folder named 'Scripts', unzip it into there and it'll be in the same place on the UI as the X/Z plot option.
(Ignore anything it has to do with lora's by the way)
is it posible to put more pictures in the gif than 12 ?
@Jaulustus If you right click on the script and click edit, you'll be able to see the lines that deal with prompting so you can make it do whatever you'd like it to (in theory).
prompts.append(f"topless, (small breasts:1.2), nipples, {character_prompt}, ") The {character_prompt} is what you're doing in SD yourself, so just copy and paste the entire line for however many frames you want, with the prompt in bold changing gradually throughout all the lines.
@BDinc worked nice.
@Jaulustus Once again, wasn't my work so I can't take credit, I only changed the prompts!
@BDinc than i wait for other version
do you mean V2 prompt still will be hardcoded?
@Tua I don't follow sorry. The second version I'll do, will be double the frames more than likely. I'll increase the prompt strength .05 each time, which will hopefully get a smoother transition. Of course with more frames, there's more opportunities for the rest of the image's details that might alter, hands/hair/face/pose, so it might not turn out as good. You'd have to manually edit the gif afterwards to speed it up slightly or the whole thing will be scuffed.
@BDinc oh, i thought there would be hardcoded prompt for pregnancy in V2
@Tua No, I've tried testing the pregnancy prompt itself, and it isn't great. The only pregnant Lora I've found is really stylised so it looks rubbish as it suddenly changes half way through the animation.
@BDinc I'm trying right now to implement Range system from X/Y/Z plot for better flexibility
after using it have some questions.
1. when selected theres lora name and lora strength, do i needa bother about it?
2. after selecting and running it with default setting, it does generate the image, but wheres the gif file located though?
Ignore all that stuff. It was originally designed to work alongside a lora that prompted deepthroating to make a gif, I've changed it to make it do breast growth instead.
I don't know enough about coding to get rid of the relevant lines that have anything to do with the lora so just leave it blank.
Once you run it with the script, after all images are generated, it should export a gif to "outputs/txt2img-images/txt2gif-Breastgrowthfast". You can change that line to have it export wherever, or if you have a prefered location for your outputs, you should change it to wherever they need to go and it should work perfectly.
(Also I'd love to see if you've made anything good with it!)
@BDinc I am currently making changes to the script in order to make it better. Would you know the license on this? (Just wanna check the permissions on it).
@silverace71 mizumizuni-concept-v1.5 was the original Lora+Script, but as someone has pointed out, the publisher has either deleted their profile or had it deleted for some reason. I don't know his username to be able to credit him properly.
I'd like to see the changes either way if you end up uploading it yourself!
@BDinc tyvm will do!
This seems really cool idea! But i think i missed something. Script execute like a charm, but at the end all the pictures are giving the same breast size, just slightly different skin tone, or shadows. Did i missed something? The sample pics, and the gif someone posted are really nice, so i'm sure i did something wrong :'D
Can you post an example image? Or even an end gif? Bear in mind it'll depend on what model you're using, if you're using a character or style lora how flexible they are etc.
@BDinc Woah, thanks for the quick reply. Seems like i can't add images, when i'm commenting, or even when i'm editing my post, but will make a new post soon, as i think i found the problem.
@Zsentl Not to worry! I can't attest to the code too much since I didn't make it as I mentioned, but if I can help at all getting it working for you I'll try! I'm currently just altering the prompt lines to make a weightloss/workout version!
your prompt must start with a smaller breast, so that the generated image will become larger.
@huzhehuan894 your prompt in Stable Diffusion should absolutely not start with smaller breast. All this script does is prompt in small>to>huge for each iteration of the image, so you shouldn't manually prompt breast side to try to clash with the script.
with this script in the scripts folder, it doesnt show up in the scripts drop-down and i get ValueError: could not convert string to float: '' in the console when trying to generate any image with no script at all.
@TheSlad errrr, no idea I'm afraid. Had you opened a fresh SD+UI after installing the script I assume?
its great working ! Thank you sharing this !
BTW: the deepthrow lora is ban because some other problem.
you can fine the author's hugging face here @aurenigma
Really enjoying using this. I'm curious if there's a way to get this consistency and resolution with something like the original deepthroat animation.
One thing I've noticed, not sure if it's something I'm doing wrong or not, but if I use openpose controlnet, it generates the pose annotator for each image and breaks the gif. I've just been using Flowframes to make the gifs, but I'm curious if there's a way to fix that.
Is there something in controlnet that allows you to NOT produce the annotator at all?
@BDinc Not that I could find, I've only been using controlnet for a bit over a week, so I'm not fully aware of everything that's possible yet. I could just be missing something.
@Hellbacon I wouldn't know where to look, or how to change it if it came down to a coding issue. It's a pain, but you might just have to assemble the gif manually with every other frame in that case.
@BDinc It's not too much of an issue doing it manually, so no biggie. Just thought I'd put it out there in case anyone has any solutions.
Very good script i fixed a little to get a better result
Okay!
@BDinc I'll send it to you if you want
@scooller If you'd like, feel free to post the results here in case anyone wants to use it? Or upload it yourself and link I'll post a link as an alternative?
Yes, please upload your tweaks! I'm currently working on a modification to allow defining of custom templates to be applied across the generation series.
Yes, it would incredibly generous of you to share your mods :)
here https://github.com/scooller/stabledifusionscript only with the name txt2gif-Breastgrowthfast
@scooller So what does it do differently exactly?
@BDinc I didn't write it, but It's just adding more generated frames with more drastic transitions. Everything between the "quotes" on each prompts.append line gets added as a prompt for another frame to be generated, and you can add as many of those prompts.append(...) lines as you want to make it generate more images.
@eisenholz Sorry he'd said he'd fixed it a little, so I assumed he'd improved the code in some way. I was asking what it does differently in terms of the script as a whole, not just changing the prompts.
@BDinc i'm remove the Lora section, add new term "flat chest" and "gigantic breasts" and add more duration to the gif and put all in one folder "txt2gif" for unification
@scooller All the individual frames, as well as the gif you mean? That is better then. The lora section is a nice option to have, in case what you're trying to do reacts to one, but I've found most of the stuff I've used it for personally doesn't use them anyway.
Where should this script be installed?
In the webui-master directory, should be a folder named scripts. Save in there. Should then be accessible from the dropdown menu where the XY plot is found
Great work, just one issue for me, the script works normally and outputs a gif just fine, but every time I applied ControlNet to it, the script will draw all the frames, but will not output a gif of them, instead it prints "AttributeError: im must have seek method", can you fix that for me? Much appreciated!
Nope, unfortunately I think you'll have to assemble to non controlnet frames separately using something like ezgif.com
@AllTheEpochs Okay, thank you very much, I just want to make sure that this issue was shared by most users or just me, thanks.
thanks! I got it to make a gif, but instructions are so unclear - do I need to write both the desired starting and ending sizes of the boobs? do i need to write in the prompt that she's flat chested, and then the script makes her bigger later on? some guidance from anyone would be great! ty
Saw mention of a V2 in other comments and was curious how that was coming along.
That said, as a BE fan, I do find myself saddened by the limits most quality models settle at for bust size. I know you also mentioned inexperience in incorporating loras relative to the old functionality, but I'm hoping either that gets figured out or a way to incorporate loras (such as hyperbreasts) within the main prompt without breaking the script, as the times I did try just put out intangible garbage.
@WeisseningBlitz to be honest, I really never got around to trying to alter it too much, as it was already made by someone else and I wasn't experienced enough to alter it that much, I only incorporated the breast enlargement as it was something the script just managed to do all by itself.
The lora functionality wasn't needed for the script to function for what I intended it to do, but it's still in there, regardless, you can just quote the lora within the prompt inside the script itself, if you're wanting to do something such as hyperbreasts.
@AllTheEpochs I'll have to give that a shot, then. Hopefully RNGesus will cooperate with transitions. XD
I remember you plan to train a pregnant like lora like this, is that plan still on the way?
I couldn't find a decent enough pregnancy lora that the script got on well with, as plain prompts just weren't cutting it.
If you find one that works from low weights up to high weights, you should just be able to edit the script and put it in to each line, so it gradually grows.
<lora:pregnancy:0.5>, <lora:pregnancy:0.6>, <lora:pregnancy:0.7>, <lora:pregnancy:0.8>, etc.
It seems to be a good lora, but I don't know how to use it like the effects below. How to use it?
It's not a lora, it's a script. Put it in the script folder and you'll see it in the drop down where XYZ grid is.
How exactly is thte gif exported? I tried it, and it made the images, but I don't know where the gif was exported, if anywhere.
Mine makes a new folder within the txt2img directory with the gif inside
How do you increase the frames? Adding prompts in the script doesn't seem to change how many images it creates.
Each new line starting with "prompts.append" will create a new image. The original script has 18 and it created 18 images, I cut it down to 2 lines and it only created 2 images.
@AllTheEpochs It looks like my script wasn't updating in 1111 unless I restart the UI. I can now edit the script, add or remove lines, restart UI and it seems to respond. Thanks for confirming that it's editable.
@gravve186 no worries! Bear in mind as well, there's a line of code here:
# Function to set title
def title(self):
return "txt2gif-Breastgrowthfast"
so if you make different versions of the script, to do different things, make sure to edit this as well as renaming the script file itself.
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.