Hi, can support be added to face editor so that each face can have different expression in group picture? Thank you

Random order is great, but sometimes I want some control over the order, so can you add an option to allow sequential order, meaning it will allow reading from the rows from first to last, something like, __lora__ will pick a random row from lora.txt, ||lora||will pick from the first row to last in sequential order?

The reason I'm asking is that I want to do side by side comparison of several character loras. I have saved these loras in lora.txt, but since the row is randomly picked, sometimes I have trouble linking character to lora used. If there is an option of sequential order, it will be straightforward.

It would be nice if it was possible to add a new file directly from the interface. Currently, you have to reuse existing files or search for the appropriate folder via SSH.

Changing the file name would also be very useful.

Yesterday the script worked just fine, and I was able to get random values from several different wildcard .txt files and all was right with the world, then last night, after I fired up stable diffusion/automatic1111 again, it just stopped working. I get an error of (color1 is just one of many files that worked fine before) in the cmd console:

WARNING:root:Could not find any wildcard files matching color1

The .txt files are in the proper \stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\wildcards directory. I have no idea what happened. Nothing changed as far as I know, I didn't install any other extensions in the time between when it worked and later when it didn't, and stable diffusion didn't update.

0
100%|████████████████████████████████████████████████████████████████████████████████| 110/110 [00:18<00:00, 5.85it/s]
2023-01-28 22:26:40,872 - root - INFO - Calibrating color correction. | 110/6666 [00:18<18:06, 6.03it/s]
Rendering animation frame 2 of 300
creating in between frame 0 tween:0.50
resized to 500x1000
creating in between frame 1 tween:1.00
resized to 500x1000
resized to 500x1000
Deforum progress: 2%|█ | 110/6666 [00:19<18:57, 5.76it/s]
Error completing request
Arguments: ('task(c4663zv6flaeiz1)', None, False, '', '3D', 300, 'wrap', '0:(0)', '0:(1.02+0.02sin(23.14*t/20))', '0:(0)', '0:(0)', '0:(1)', '0:(0)', '0:(0)', '0:(0)', False, '0:(0)', '0:(t%15)', '0:(0)', '0:(53)', '0: (0.02)', '0: (0.6)', '0: (1.0)', '0: (7)', False, '0: (25)', '0: (40)', '0: (200)', '0: (10000)', '0: (t%4294967293)', False, '0: ("Euler a")', '0: ("!({everywhere}^({init_mask}|{video_mask}) ) ")', False, '0: ("!({everywhere}^({init_mask}|{video_mask}) ) ")', False, '0: ("model1.ckpt"), 100: ("model2.ckpt")', '0: (5)', '0: (1.0)', '0: (0.1)', '0: (0.0)', True, 'Match Frame 0 LAB', 1.0, 2.0, 'perlin', 8.0, 8.0, 4, 0.5, True, 0.3, 200.0, 10000.0, 40.0, 'border', 'bicubic', False, '/content/video_in.mp4', 1, 0, -1, False, False, '/content/video_in.mp4', False, 4, False, '20220829210106', ong neck"\n}', 512, 1024, False, False, True, 0, 0, -1, 'Euler a', False, 1, 0.01, 0, 0, 110, 0.0, 1, False, 1, True, True, False, False, False, False, 'Deforum', '{timestring}{index}{prompt}.png', 'iter', 1, False, False, True, 0.6, 'https://user-images.githubusercontent.com/14872007/195867706-d067cdc6-28cd-450b-a61e-55e25bc67010.png', False, False, False, True, 'https://www.filterforge.com/wiki/images/archive/b/b7/20080927223728%21Polygonal_gradient_thumb.jpg', 1.0, 1.0, 5.0, 1, True, 4, 'reroll', False, 28.0, 'FFMPEG mp4', 'C:\stable-diffusion-webui\venv\lib\site-packages\imageio_ffmpeg\binaries\ffmpeg-win64-v4.2.2.exe', 17.0, 'veryslow', 'None', 'snowfall.mp3', False, False, 300.0, 'x0_pred', '/content/drive/MyDrive/AI/StableDiffusion/2022-09/20220903000939_%05d.png', '/content/drive/MyDrive/AI/StableDiffusion/content/drive/MyDrive/AI/StableDiffusion/2022-09/kabachuha/2022-09/20220903000939.mp4', False, 'RIFE v4.6', 'x3', 'x2', False, '', True, False, 'None', False, 'None', 'Farneback', False, 'None', False, 'None', False, False, '0:(1)', '0:(0.5)', '0:(1)', '0:(100)', '0:(0)', False, '{\n "0": "https://cdn.discordapp.com/attachments/1047760262637297664/1049851726200578058/a_dragon_flying_over_a_futuristic_city_dramatic_sky_cinematic_view_seed-1703716904ts-1670374556_idx-0.png",\n "50": "https://cdn.discordapp.com/attachments/1047760262637297664/1049901219004555274/a_digital_painting_of_a_fox_standing_in_the_grass_on_a_hill__trees_beh_t_outpaint_from-down_seed-1205071278ts-1670386356_idx-0.png",\n "100": "https://cdn.discordapp.com/attachments/745415989633482882/1048084925359857704/00024-2977479109-wings_hubble_images_jwst_images_6.png",\n "150": "https://cdn.discordapp.com/attachments/1011355905490694355/1047133257097031680/31_0031_31.jpg",\n "200": "https://cdn.discordapp.com/attachments/1047760262637297664/1049851726200578058/a_dragon_flying_over_a_futuristic_city_dramatic_sky_cinematic_view_seed-1703716904ts-1670374556_idx-0.png"\n}', '0:(.75)', '0:(.35)', '0:(.25)', '0:(20)', '0:(.075)', '

Deforum v0.5-webui-beta

\n
• The code for this extension: Fork of deforum for auto1111's webui.
\n
• Join the official Deforum Discord to share your creations and suggestions.
\n
• For general usage, see the User guide for Deforum v0.5.
\n
• For advanced animations, see the Math keyframing explanation.
\n
• Alternatively, use sd-parseq as a UI to define your animation schedules (see the Parseq section in the Keyframes tab).
\n
• framesync.xyz is also a good option, it makes compact math formulae for Deforum keyframes by selecting various waveforms.
\n
• The other site allows for making keyframes using interactive splines and Bezier curves (select Disco output format).
\n
• After the 2022-12-30 update, the default noise type is Perlin noise due to its great frame coherence improvements. If you want to use the old noise and replicate the previous settings, set the type to "uniform" in the Keyframes tab.
\n

Import settings from file

Animation settings

Motion parameters:

2D and 3D settings

Prespective flip — Low VRAM pseudo-3D mode:

Coherence:

3D Depth Warping:

Video Input:

Interpolation (turned off atm)

Resume animation:

Prompts

animation_mode: None batches on list of prompts. (Batch mode disabled atm, only animation_prompts are working)

Important change from vanilla Deforum!

This script uses the built-in webui weighting settings.

So if you want to use math functions as prompt weights,

keep the values above zero in both parts

Negative prompt part can be specified with --neg

Run settings

Sampling settings

3D Fov settings:

Batch settings

Init settings

3D settings

Steps scheduling:

Video output settings

To enable seed schedule select seed behavior — 'schedule'

\n Note that parseq overrides:\n

\n
• Run: seed, subseed, subseed strength.
\n
• Keyframes: generation settings (noise, strength, contrast, scale).
\n
• Keyframes: motion parameters for 2D and 3D (angle, zoom, translation, rotation, perspective flip).
\n

\n
• Run: Sampler, W, H, Restore faces, tiling, highres fix, resize seed.
\n
• Keyframes: animation settings (animation mode, max_frames, border)
\n
• Keyframes: coherence (color coherence & diffusion cadence)
\n
• Keyframes: depth warping
\n
• Video output settings: all settings (including fps and max frames)
\n

Anti-blur settings

Perlin noise params, if selected.

Hybrid Video Compositing in 2D/3D Modeby reallybigname

• Composite video with previous frame init image in 2D or 3D animation_mode (not for Video Input mode)
• Uses your Init settings for video_init_path, extract_nth_frame, overwrite_extracted_frames
• In Keyframes tab, you can also set color_coherence = 'Video Input'
• color_coherence_video_every_N_frames lets you only match every N frames
• Color coherence may be used with hybrid composite off, to just use video color.
• Hybrid motion may be used with hybrid composite off, to just use video motion.

• The alpha schedule controls overall alpha for video mix, whether using a composite mask or not.
• The hybrid_comp_mask_blend_alpha_schedule only affects the 'Blend' hybrid_comp_mask_type.
• Mask contrast schedule is from 0-255. Normal is 1. Affects all masks.
• Autocontrast low/high cutoff schedules 0-100. Low 0 High 100 is full range.
  (hybrid_comp_mask_auto_contrast must be enabled)

Hybrid Video Schedules

\n Supported engines:\n

\n
• RIFE v4.6, v4.3 and v4.0. Recommended for now: v4.6.
\n
• RIFE v2.3 and other interpolation engines might come in the future.
\n

\n Important notes:\n

\n
• ffmpeg has to be installed for this feature to work properly. If you can't have it, make sure to check the "keep_imgs" tab so that your interpolated frames are saved into HD'\n
• Frame Interpolation will not run if 'store_frames_in_ram' is enabled.
\n
• Audio (if provided) will be transferred to the interpolated video even if Slow-Mo is enabled.
\n
• Frame Interpolation will always save an .mp4 video even if you used GIF for the raw video.
\n

Hello, for a while I've had problems sometimes when using several wildcards with LoRAs at the same time despite lowering wildcard weights accordingly.
I have noticed that that only modifies the weight of trigger words but not of the LoRAs themselves.
I would like to ask if it would be possible to add that function.
EG: if lora:baccon:0.8 is in a wildcard file set in prompt to :0.5, then it should be lora:baccon:0.4 or lora:baccon:0.8:0,5 (idk if this last one would work).

Edit: I have just noticed that for wildcard lines with more than one word, the weight is only added to the last one. This limits greatly what can be done with the wildcard tool.
So what happens is (a, b, c)*0.4=a, b, c:0.4 instead of a:0.4, b:0.4, c:0.4

1, if I put something in the prompt area, like cube, I get a message in the command window: cube.txt not found for the cube wildcard.

2, typing artists (for example) used to pop up an auto complete. But it doesn't anymore. Possibly related, the --deepdanbooru auto complete doesn't seem to come up either.

Add support for these kinds of prompts

a very cute ((__color__ __{cat;dog;monkey;fish}__))

This is meant to look for words in a file called color.txt ofcourse as it does by default, but also when it encounters a "chunk" surrounded by __{ and }__ it should directly pick a word from in-between those

Here is some basic code that allows for this, but it could definitely be improved

def is_inline_wilcard(self, text):
	return "}" in text and "}" in text

def replace_wildcard(self, text):
	if " " in text or len(text) == 0:
		return text
		
	# very basic but it works
	if self.is_inline_wilcard(text):
		choices = text.replace("{", "").replace("}", "").split(";")
		assert len(choices) > 0
		return random.choice(choices).replace("_", " ")
	else:
		file_dir = os.path.dirname(os.path.realpath("__file__"))
		replacement_file = os.path.join(file_dir, f"scripts\\wildcards\\{text}.txt")
		if os.path.exists(replacement_file):
			with open(replacement_file, encoding="utf8") as f:
				return random.choice(f.read().splitlines())
		else:
			if replacement_file not in warned_about_files:
				print(f"File {replacement_file} not found for the __{text}__ wildcard.", file=sys.stderr)
				warned_about_files[replacement_file] = 1

	return text

I cannot find a clear tutorial on this Wildcards thing, besides some cool stuffs people wrote about this feature at Reddit.

I did try all the steps:

• dynamic prompting
• wildcard manager set to parrotzone
• actually use the blah in the prompt

For example:
"portrait of cute lady with marvelous color hair"

I activated "Write Prompts To File" as well, hoping I could get a list of all the prompts generated via wildcards modifier, however none seems to be working?

When the seed is fixed, only the first instance is random, and the same value is used from the second instance onward.
Is it possible to have a fixed seed while generating a random value every time?

Using Send To img2img results in the loss of some parameters auto-propagating the form. Such as size, cfg, prompt and seed.

how do you install this, can you add instructions?

If the file uses LF as the line seperator, plugin works well.If the file uses CRLF(default on Windows), only replace wildcards to the last line.

A wildcard added to the negative prompt is ignored.

I've looked everywhere & can't find a workaround. Is there any way you could update the extension to include the ability to use wildcards for negative prompts as well?

Looks like this script didn't bring over the updates I made to the existing wildcard script that handled image grid generation as a final step, fixed seed issues, and added wildcard prompt info to the PNG metadata.

If those aren't something to be handled by extensions anymore then perhaps this can be closed, but otherwise these two commits should be simple enough to port to the new script. Let me know if you'd prefer if I submitted as a PR, or if there was a reason the outdated script was used (these were previously already merged into the jtkelm2 script).

AUTOMATIC1111/stable-diffusion-webui@8969bb4
AUTOMATIC1111/stable-diffusion-webui@5a4a343
(note that the 2nd commit fixes a few issues noticed/mentioned when developing 1st commit)

Thanks,
Aphix

I don’t know if there is a way to turn off the random function? Sometimes I want to test the prompt words in order.

Hires prompts are not properly modified to remove __wildcard__ tokens. Stepping through the debugger in sample_hr_pass in processing.py in the main stable-diffusion repo, I see that a number of hires-related prompt variables do not have the wildcards properly replaced: at least self.all_hr_prompts, self.hr_prompts, and self.hr_prompt. Anecdotally, this explains why all of my generations look decent on the first txt2img pass, but become very same-y after the hires pass.

I have to assume this was broken a long time ago when the base repo introduced the ability to set a different prompt for the hires pass AUTOMATIC1111/stable-diffusion-webui@ff0e171

The dynamic prompts extension had to put in a fix for this, so I assume it's also needed here adieyal/sd-dynamic-prompts#474

I moved all my wildcards over to /extensions/scripts/wildcards since the move to an extension but it's still looking for them in /scripts/wildcards/

If i move them back to /scripts/wildcards/ then it works again and loads them all in as available.

When using the Combinatorial generation, is there any possibility to use the used keyword as a token to use in the filename?
So that it would be possible to save images with the same word(s) as or in the filename?

I don't think it's correctly handling loras in wildcard prompts in a batch

I have a wildcard file of the following lines:

lora:TLOK_Asami_Sato_v4:0.6, TLOK_Asami_Sato, \green eyes, black hair, main_outfit
lora:YangXiaoLong:1, ((yellow hair)), yang xiao long, (prosthetic arm), smile, ahoge purple eyes, brown jacket, waist coat
lora:nora_valkyrie:0.6, nora_valkyrie, looking at viewer, smile, short hair, [[blue eyes]], gloves, jacket, pink skirt, pink belt, short sleeves, puffy sleeves, fingerless gloves, orange hair

They are character loras for Asami Sato from Legend of Korra, and Yang Xiao Long + Nora Valkyrie from RWBY, for visual reference. When I run this prompt using a batch size of 1 and batch count of 6, I get what I'd expect:

But when I generate with a batch size of 6 and batch count of 1, I think it's including the lora used for the first image in all the images generated.

Here's the one that starts with Nora:

Here's the one that starts with Yang:

Here's the one that starts with Asami:

Till yesterday everything worked fine, but today, using the same wildcards and prompts it doesn't works and prints this in the terminal each time i start a generation that includes wildcards in it's prompt:

*** Error running process: D:\Red-AI\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.py
    Traceback (most recent call last):
      File "D:\Red-AI\stable-diffusion-webui\modules\scripts.py", line 718, in process
        script.process(p, *script_args)
      File "D:\Red-AI\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.py", line 49, in process
        p.all_prompts = self.replace_prompts(p.all_prompts, p.all_seeds)
      File "D:\Red-AI\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.py", line 42, in replace_prompts
        res.append("".join(self.replace_wildcard(chunk, gen) for chunk in text.split("__")))
      File "D:\Red-AI\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.py", line 42, in <genexpr>
        res.append("".join(self.replace_wildcard(chunk, gen) for chunk in text.split("__")))
      File "D:\Red-AI\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.py", line 27, in replace_wildcard
        return gen.choice(f.read().splitlines())
      File "C:\Users\enric\AppData\Local\Programs\Python\Python310\lib\codecs.py", line 322, in decode
        (result, consumed) = self._buffer_decode(data, self.errors, final)
    UnicodeDecodeError: 'utf-8' codec can't decode byte 0xe3 in position 7: invalid continuation byte

The wildcards extension is great except for one thing; it often repeats the same token multiple times in a prompt. If there was a way to prevent a wildcard from repeating within the same prompt it would be quite helpful.

File O:\AI\SynologyDrive\extensions\stable-diffusion-webui-wildcards\wildcards|.txt not found for the | wildcard.

"|.txt" This text file does not exist.

It is making my images black and white. I never asked this to happen. I don't see a setting to turn it off.

Edit: oops, wrong plugin-I was using the wildcards in dynamic prompts. Please delete this.

What is the reason for this?
I can't use the wildcard color to input random words in forward words, but I can use the wildcard color to input random words in reverse words.

There is only one case that can be used in the forward word, that is, only color without any words, and just adding a space or comma after color will not work for random input words.

Hi,

looks like there is a problem with how it adds itself to the ui. In a1111 clone https://github.com/vladmandic/automatic it shows up twice in the tabs, and also the dropdown menus at the bottom show twice in each tab. Didn't try with other a1111 clones.

The output image does reflect recognizing wildcard strings with the double underscore wrapping.

suggestion -- use left click to add catagory and individual items to prompt rather than having to copy on paste each -- or control click shortcut to add just an idea to speed up things if it has not been suggested before

Hi, I'm trying this plugin for the first time, I created the name.txt in the wildcards folder in the extensions but when I try and call the wildcard in the prompt I am met with the following error. I am using __name__ in my prompt as directed. I have tried different names for the wildcards also.

WARNING:root:Could not find any wildcard files matching name
INFO:dynamic_prompting.py:Prompt matrix will create 1 images in a total of 1 batches.

image generation is normal but without the wildcards of course.

I have tried placing the text file in various locations including the root directory and in a directory called wildcards in the root and could not figure it out.
Thanks so much

I got following error when trying to use a list of names from this link (https://civitai.com/models/70708?modelVersionId=79565) and saved in a txt file in wildcards folder.
Error running process: D:\stable-diffusion\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.pys] Traceback (most recent call last): File "D:\stable-diffusion\stable-diffusion-webui\modules\scripts.py", line 418, in process script.process(p, *script_args) File "D:\stable-diffusion\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.py", line 41, in process prompt = "".join(self.replace_wildcard(chunk, gen) for chunk in prompt.split("__")) File "D:\stable-diffusion\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.py", line 41, in <genexpr> prompt = "".join(self.replace_wildcard(chunk, gen) for chunk in prompt.split("__")) File "D:\stable-diffusion\stable-diffusion-webui\extensions\stable-diffusion-webui-wildcards\scripts\wildcards.py", line 25, in replace_wildcard return gen.choice(f.read().splitlines()) File "C:\Users\Meng Lu\AppData\Local\Programs\Python\Python310\lib\codecs.py", line 322, in decode (result, consumed) = self._buffer_decode(data, self.errors, final) UnicodeDecodeError: 'utf-8' codec can't decode byte 0xe9 in position 221: invalid continuation byte

having this error