[Bug]: can't anymore enter custom list as text in 'checkpoints name' field in 'x/y/z plot'

[DiegoRRR]

Is there an existing issue for this?

• I have searched the existing issues and checked the recent builds/commits

What happened?

I just updated, and I have an issue with the new way the "Checkpoint name" field in "x/y/z plot" works.
See I have several custom lists of checkpoints, at text with a coma between each of them, for when I want to compare either backgrounds, or portrait paintings, and so on, and I used to enter one of these lists in the field depending on what I want to do. I am sure many other people were doing this way, because we all end up having tons of checkpoints and we don't want to always compare all of them.
But now I can only add checkpoints one by one or click the book icon to add all of them.
I can't anymore enter one of my custom lists as text like before. :(
One should not exclude the other though. It should still be possible to enter a list as text, and it would just add every checkpoint of the list. There is no reason for the new system to be incompatible with text input. Please fix this.

Steps to reproduce the problem

1. Go to 'x/y/z plot'
2. Set y as 'checkpoints name'
3. enter a list, for example : "paint 526sCuspOfSerenity_v1.safetensors, paint aiceberg 15.safetensors, paint animatrix_v13.safetensors, paint artErosAerosATribute aerosPruned.safetensors, paint artisantheory v1.safetensors, paint artMNX M .safetensors"

Result: text input is greyed then gets deleted

What should have happened?

Every checkpoint in the list should be added

Commit where the problem happens

89f9faa

What platforms do you use to access the UI ?

Windows

What browsers do you use to access the UI ?

Mozilla Firefox

Command Line Arguments

set COMMANDLINE_ARGS=--xformers

List of extensions

Stable-Diffusion-Webui-Civitai-Helper https://github.com/butaixianran/Stable-Diffusion-Webui-Civitai-Helper fcf5a27a (Fri Apr 14 05:39:01 2023) unknown
model_compare_script https://huggingface.co/Froddan/model_compare_script ee22f785 (Sun Dec 4 12:06:31 2022) unknown
multi-subject-render https://github.com/Extraltodeus/multi-subject-render cc9d2340 (Tue Nov 29 00:22:31 2022) unknown
sd-dynamic-prompts https://github.com/adieyal/sd-dynamic-prompts 601871bb (Sat Dec 24 17:35:52 2022) unknown
sd-webui-controlnet https://github.com/Mikubill/sd-webui-controlnet fc9e4d61 (Fri Feb 17 16:47:17 2023) unknown
sd_smartprocess https://github.com/d8ahazard/sd_smartprocess 5fa74185 (Thu Dec 15 19:48:36 2022) unknown
stable-diffusion-webui-aesthetic-gradients https://github.com/AUTOMATIC1111/stable-diffusion-webui-aesthetic-gradients 411889ca (Sat Oct 22 13:03:10 2022) unknown
stable-diffusion-webui-model-toolkit https://github.com/arenatemp/stable-diffusion-webui-model-toolkit 345923d0 (Wed Jan 25 16:16:00 2023) unknown
ultimate-upscale-for-automatic1111 https://github.com/Coyote-A/ultimate-upscale-for-automatic1111.git 0a3d03a4 (Tue Feb 7 06:07:23 2023) unknown

Console logs

fatal: unable to access 'https://github.com/AUTOM
.git/': Failed to connect to github.com port 443
venv "D:\apps\AI\stable-diffusion-webui__automati
Python 3.10.6 (main, Sep 19 2022, 15:45:37) [MSC
Version: v1.2.1
Commit hash: 89f9faa63388756314e8a1d96cf86bf5e066
Installing requirements

Installing Send2Trash==1.8.0

Installing requirement for sd-webui-controlnet

loading Smart Crop reqs from D:\apps\AI\stable-di
tensions\sd_smartprocess\requirements.txt
Checking Smart Crop requirements.

Launching Web UI with arguments: --xformers
=================================================
You are running torch 1.12.1+cu113.
The program is tested to work with torch 2.0.0.
To reinstall the desired version, run with comman
Beware that this will cause a lot of large files
there are reports of issues with training tab on

Use --skip-version-check commandline argument to
=================================================
=================================================
=
You are running xformers 0.0.14.dev.
The program is tested to work with xformers 0.0.1
To reinstall the desired version, run with comman
.

Use --skip-version-check commandline argument to
=================================================
=
Civitai Helper: Get Custom Model Folder
Civitai Helper: Load setting from: D:\apps\AI\sta
111\extensions\Stable-Diffusion-Webui-Civitai-Hel
Civitai Helper: No setting file, use default
Error loading script: simple_depthmap.py
Traceback (most recent call last):
  File "D:\apps\AI\stable-diffusion-webui__automa
ne 257, in load_scripts
    script_module = script_loading.load_module(sc
  File "D:\apps\AI\stable-diffusion-webui__automa
py", line 11, in load_module
    module_spec.loader.exec_module(module)
  File "<frozen importlib._bootstrap_external>",
  File "<frozen importlib._bootstrap>", line 241,
  File "D:\apps\AI\stable-diffusion-webui__automa
t-render\scripts\simple_depthmap.py", line 11, in
    from repositories.midas.midas.dpt_depth impor
  File "D:\apps\AI\stable-diffusion-webui__automa
s\dpt_depth.py", line 5, in <module>
    from .blocks import (
  File "D:\apps\AI\stable-diffusion-webui__automa
s\blocks.py", line 21, in <module>
    from .backbones.next_vit import (
  File "D:\apps\AI\stable-diffusion-webui__automa
s\backbones\next_vit.py", line 8, in <module>
    file = open("./externals/Next_ViT/classificat
FileNotFoundError: [Errno 2] No such file or dire
assification/nextvit.py'

Loading weights [a5ef43eb16] from D:\apps\AI\stab
11\models\Stable-diffusion\photo uberRealisticPor
Creating model from config: D:\apps\AI\stable-dif
figs\v1-inference.yaml
LatentDiffusion: Running in eps-prediction mode
DiffusionWrapper has 859.52 M params.
Running on local URL:  http://127.0.0.1:7860

To create a public link, set `share=True` in `lau
Startup time: 27.5s (import torch: 6.8s, import g
other imports: 1.5s, list SD models: 0.3s, setup
 11.4s, create ui: 4.4s, gradio launch: 0.2s).
Loading VAE weights specified in settings: D:\app
tomatic1111\models\VAE\vae-ft-mse-840000-ema-prun
Applying xformers cross attention optimization.
Textual inversion embeddings loaded(0):
Textual inversion embeddings skipped(1): SDA768
Model loaded in 10.7s (load weights from disk: 1.
eights to model: 1.4s, apply half(): 1.4s, load V
 0.7s, load textual inversion embeddings: 1.2s).

Additional information

No response

[Sakura-Luna]

Textbox and dropdown are two different controls. Adding the latter has made the code logic a lot more complicated. It is feasible to realize your needs, but it will make the code more difficult to maintain.

Maybe sometime Gradio will expand its capabilities and not be as rudimentary as it is now.

[w-e-w]

I also noticed this issue when the drop down thing was introduced

the new doctor allows for easier setup
but the old comma separated list allows for easier importing of old configuration (after it has been properly configured)

the new drop down also created a second issue
in that the list becomes very long

this makes the "field available values into field" button effectively useless in some cases

I was thinking maybe we can just create a second input checkpoint axle but for comma separate list
users would just choose which one to use
axle checkpoint checkpoint (comma)

[DiegoRRR]

Maybe a simple and quick solution is to add a "txt" button ("paste list as text") next to the book button ("Paste available values into the field"). Clicking that button makes a textbox appear below the dropdown, with a "add" button next to it. We enter our text list in it, click "add", and voila.
Or just bring back the old one as a second axle, like w-e-w said call it "checkpoint (comma)". Please someone do that.

[Mistborn-First-Era]

I like DiegoRRR's idea of adding a second axle for CSV lists.
I also have the problem that sometimes I cannot open the dropdown list again if I forget to select 1. Then I have to go through the whole list again and select my 20 or so models for testing.

[Sakura-Luna]

Whether it is unhiding the textbox or adding new controls, it is easy to cause new problems, especially now that there are already problems, thanks to Gradio's crudeness, there will be more such problems in the future.
If many people want to get back the original textbox, the best option is to roll back the dropdown.

[huyzer]

The new way is NO good. Please bring back the old way with comma separated lists.

[tbrebant]

Upvoting that too.

[ljleb]

Another issue I ran into related to the dropdown is that when I merge models locally using sd-meh, the sha256 changes. whenever I restart the webui server, I also have to reload the front end so that the hashes in the names in the list are updated. This is not cool when I have to pick the same 5 models and reconfigure the ui every time.

[w-e-w]

Another issue I ran into related to the dropdown is that when I merge models locally using sd-meh, the sha256 changes. whenever I restart the webui server, I also have to reload the front end so that the hashes in the names in the list are updated. This is not cool when I have to pick the same 5 models and reconfigure the ui every time.

for this use case probably not including the hush in the drop-down is better
the path should be sufficient I guess

but no hush serves as a universal identifier
which would be kind of important after the fact

maybe an option to have hash or no hash
or maybe specifying using only the path but the hash would be appended to the lables

[ljleb]

but no hush serves as a universal identifier

Maybe it could be possible to update the list when loading a model then? Or just manually. IIRC it should be possible to update the possible values even after creating the UI. Maybe put a realod button next to the dictionary icon?

[ljleb]

In any case this is a different issue I think. I can open a different issue, maybe I'll propose changes if I ever find the time.

[w-e-w]

made a PR to restore CSV functionality
you can toggle the behavior from a checkbox
if anyone has time please test

[w-e-w]

PR merged into dev