rvion / cushystudio Goto Github PK

View Code? Open in Web Editor NEW

570.0 13.0 39.0 68.34 MB

🛋 The AI and Generative Art platform for everyone

Home Page: https://docs.cushystudio.com/

License: GNU Affero General Public License v3.0

TypeScript 93.81% HTML 0.29% CSS 1.32% JavaScript 4.01% Batchfile 0.33% Shell 0.24%

ai comfyui cross-platform generative-ai generative-art live-coding

cushystudio's Introduction

🛋️ CushyStudio

The Generative AI Platform for everyone

CushyStudio is the go-to platform for easy generative AI use, empowering creatives of any level to effortlessly create stunning images, videos, and 3D models.

What You'll Love:

CushyApps: A collection of visual tools tailored for different artistic tasks. Simplify your AI art creation process and have fun exploring a wide range of versatile to niche tools in the Cushy Library.
CushyKit: Speed up your projects with our extensive toolkit for custom apps development and task automation. Quickly design interfaces, add custom logic, and access tools like ComfyUI in a user-friendly, preconfigured typescript environment.

Dive into the AI Revolution with CushyStudio. Unleash your creativity, share your projects, and connect with a vibrant community pushing the boundries of generative AI. Experience the fusion of simplicity and innovation.

Warning

⚠️ CushyStudio is under active development, so you might encounter incomplete features or issues due to rapid updates. Your feedback is vital for us to refine and expand the project. Please report any critical problems you find in our discord or open a GitHub issue.

QuickStart guide:

Have Git installed

Clone this repository with git

git clone https://github.com/rvion/CushyStudio
cd CushyStudio

Install with _windows_INSTALL.bat or _mac-linux-install.sh
Start with _mac-linux-start.sh or _windows-start.bat

Screenshots and Videos

🛋 CushyStudio

Getting Started

Going further

Community

cushystudio's People

Contributors

Stargazers

Watchers

Forkers

gitbenxing luomor-ai nakedlitttlezombie navezjt bizai3000 noellealarie zcfrank1st vaibhav-009 alexdorian7 manavb2214 jags111 kerwinchina couldnt-find-good-name jmanhype ricklove cmd-rm-rf freeo hengle berkayguven gitwilburn ax2l gggdomi ad-astra-video aiisoft kustomzone allwavemedia muou55555 birdddev asilbalaban irldoggo alexlevy0 mexicanamerican 760427741 keilladraconis v6p pengguanjun keyzf fjkiani linecode

cushystudio's Issues

Auto-complete popup should not appear immediately upon the prompt entry gaining focus

Hanging on `await flow.PROMPT()`

Using rvion.cushystudio-0.0.99.vsix and the current example given in the README, I'm not getting any output past the await flow.PROMPT() line (e.g. adding flow.showHTMLContent('TEST') to the next line does nothing). I'm not seeing any images either, although I do see them being generated in the ComfyUI output directory.

Seems like the completion event for the image generation isn't being sent/received correctly?

Here's the output log when it's hanging:

ℹ️ found WORKFLOW demo-1
ℹ️ 🔥 seed: 19040996
ℹ️ prompt requested
ℹ️ checkpoint:{"1":{"class_type":"CheckpointLoaderSimple","inputs":{"ckpt_name":"deliberate_v2.safetensors"}},"2":{"class_type":"EmptyLatentImage","inputs":{"width":512,"height":512,"batch_size":1}},"3":{"class_type":"CLIPTextEncode","inputs":{"text":"masterpiece, (chair:1.3)","clip":["1",1]}},"4":{"class_type":"CLIPTextEncode","inputs":{"text":"","clip":["1",1]}},"5":{"class_type":"KSampler","inputs":{"model":["1",0],"seed":19040996,"steps":20,"cfg":10,"sampler_name":"euler","scheduler":"normal","positive":["3",0],"negative":["4",0],"latent_image":["2",0],"denoise":0.8}},"6":{"class_type":"VAEDecode","inputs":{"samples":["5",0],"vae":["1",2]}},"7":{"class_type":"SaveImage","inputs":{"images":["6",0],"filename_prefix":"ComfyUI"}}}
ℹ️ sending prompt to http://localhost:8188/prompt

Let me know if I can help debug in any way, I'd love to get this project working for me and love what you're doing here!

Type definitions are not loaded on startup

When opening a cushy workspace, the types won't be correctly resolved until the cushy.d.ts file is manually open

vmware_6XEV7mSRMT.mp4

Simple txt2img workflow fails to parse

hamburger.json

Please see the attached hamburger workflow.

Cannot read properties of undefined (reading 'input')

Works totally fine in ComfyUI

demos library insert minified demos in the editor

Suggestion: display a tooltip with the relevant file path when hovering special tokens in the prompt box (embeddings, lora, wildcard, ...) with the mouse

Super useful, more productivity, fuck yea. Knock two birds with one stone by making it so holding SHIFT turns them clickable to open the directory. A power user's delight 👌

default models inside a map give a badly written path name back

default models inside a map give the path of file back with a \ and not a / or \\

Optionally Show preview full size in the last image panel

A small Toggle at the top of LastImage to choose if one want to see the latent preview as an image

Typo Error in README.md

In the file, there is a typographical error in the following line:

The attribute name "tyle" appears to be a typo. It should be corrected to "style".

The line - "<tr tyle='vertical-align:top'>" must be corrected to : -

"<tr style='vertical-align:top'>"

handle flow termination

success => spinning forever

Auto-complete popup does not disappear upon prompt entry losing focus

Add hover tooltips to each toolbar menu items

Hover tooltip on all of these which also contains information about what you can do with each of the functionality or window.

For example the Civitai tooltip can specify that it's simply offered as a shortcut to the civitai website and it doesn't offer any fancy integration or features other than that. Thus these tooltips help to cut down on confusion as well as better communicate the capabilities and usage patterns of the application.

Note: the tooltip should be proper minimal tooltips, not the distracting popup balloon as we currently have with the Actions button. I also suggest replacing that ballon with the proper tooltip suggested here.

Marketplace Submission: portrait emotion change

Comissioned by @rvion

https://github.com/featherice/cushy-actions/blob/main/portrait%20emotion%20change.ts

Can't start dev server

Failed to load resource: the server responded with a status of 404 (Not Found)
Logger.ts:54 [INFO] monaco ready
Logger.ts:54 [INFO] typescript ready (4.5.5)
chunk-RCPA6UVN.js:1 Uncaught (in promise) TypeError: window.TAURI_IPC is not a function
at chunk-RCPA6UVN.js:1:542
at new Promise ()
at c (chunk-RCPA6UVN.js:1:397)
at a (chunk-HNLFKTAJ.js:1:68)
at Module.o (chunk-J5WVX3SY.js:3:29)
at CushyStudio.CREATE (CushyStudio.ts:23:49)
at AppUI.tsx:22:42
at mountMemo (react-dom.development.js:17225:19)
at Object.useMemo (react-dom.development.js:17670:16)
at useMemo (react.development.js:1650:21)
DevTools failed to load source map: Could not load content for chrome-extension://cfhdojbkjhnklbpkdaibdccddilifddb/browser-polyfill.js.map: System error: net::ERR_FILE_NOT_FOUND

A way to edit run names

Would be great to have a way to rename the runs, instead of "run1,run2,run3" we could have a function in the code to rename this to whatever we like ! this way we can organize test runs for example "testing_x_thing_n" or "doing_cats" inside the project folder.

too many popups

Marketplace Submission: Vinsi.ts

https://github.com/VinsiGit/Cushy_Action

Typo Error in README.md

In the file, there is a typographical error in the following line:

The attribute name "tyle" appears to be a typo. It should be corrected to "style".

The line - "<tr tyle='vertical-align:top'>" must be corrected to : -

"<tr style='vertical-align:top'>"

Allow opening workflows as tab by drag n drop from externally

Let's say I have an old output from ComfyUI and I want to load the workflow in CushiStudio to work with it. Currently I have to copy the image to my Cushy library, but what if instead I could simply drag and drop into the application and it opens the tab?

serve the front in prod

only works during dev for now => not usable by users

Marketplace Submission: cushy-avatar-maker

https://github.com/noellealarie/cushy-avatar-maker

Restore the parameters of a previous run by drag-n-drop from gallery

In ComfyUI I have a gallery plugin and it's possible to drag an image onto the canvas to load the workflow back in, super useful. Can't seem to find anything like that in CushyStudio.

delete all legacy tauri / electron / web releases

CTRL+ENTER inputs a line break into text boxes

Running the action with CTRL+ENTER will enter a linebreak if a textbox is focused e.g. the prompt.

copy paste rich prompt items (like lora nodes) both within a rich-prompt and across rich prompts

Providing functionality that supports backend operations on the cloud.

Many individuals utilize ComfyUI in the cloud, for example, Google Colab. In my opinion, if CushyStudio could also support this backend, it would enable more people to use this tool.

doc issue?

I ran

git clone https://github.com/rvion/CushyStudio
cd CushyStudio
npm install
npm start

And I got this:

Are there some missing instructions, e.g. "Install ComfyUI first"?
N00bs gonna n00b.

Support typescript / types in the code editor

https://stackoverflow.com/questions/55132485/how-can-i-get-the-transpiled-code-from-the-monaco-editor

libssl.1.1 error

Flysoft-Studio/magenta@2415e82

Watch in the background for newly added Apps in the library

Currently we have to reload the application

Auto-generated tab names are confusing

Some tabs selected seemingly at random are renamed to a random color word. I don't really get it and it's confusing ...

More issues template / rename issue templates

Efficiency nodes optional pin not seen as optional in cushystudio

the new update for efficiency nodes added a script pin to the ksampler efficient node and added a script node for the pin to make an x/y plot, when using the sampler node (efficient) it errors out and sees the script bit as required instead of optional

Incorrect filesystem paths for vscode file URIs

Issue

Import flow fails on non-nix systems.

Other functionality may be impacted, this was the only place immediately apparent.

Solution

In multiple places, vscode.Uri.file(<path>).path is used to get the filesystem path; however, vscode.Uri.file(<path>).fsPath should be used instead.

Example: Workspace#importCurrentFileAsComfyUIPNG uses
const relPathStr = vscode.Uri.file(absPath).path.replace(this.wspUri.fsPath, '.'). The this.wspUri.fsPath is correct; however, vscode.Uri.file(absPath).path is not. This causes the replace to fail and thus the import to fail. Switching to vscode.Uri.file(absPath).fsPath fixes this issue.

The same issue is present in Workspace#importCurentFileAsComfyJSON.

Flow Summary and Image Output Area Empty

Currently the output window goes completely blank on every run. Once I close the pane, and re-run the workflow, it flashes some output for a moment and then goes blank until I close it again. This feels possibly related to the flow summary as this behavior used to only happen when I attempted to generate a flow summary.

Suggestion: rename 'Actions' button to 'Library'

I think Actions is too vague, it's not clear which actions we are talking about as it could also be the actions I already have open i.e. my tabs. Library on the other hand connects it to the folder and makes it clear that it's the template not yet loaded into a buffer to use it.

Consider creating a stable branch

In the past I tried CushyStudio a month or two and gave up on it because I couldn't seem to get ahold of a working commit. Right now I am trying it again and wow it's impressive, keep up the great work folks, finally we are starting to get real production-grade software. However even though it worked fine, certain things seem very WIP or bleeding edge, so I think a stable branch could could save some superfluous tickets and improve the adoption further!

UI inconsistencies

The name "mask" is push down for some reason.
The image selectors are not the same if they are a group or not are Opt or not.

Design concerns about the `ask*` API

Right now it seems that we're queuing a "step" for each ask request and popping them in LIFO fashion
This makes the following workflow behave rather unexpectingly:

WORKFLOW('multi-ask', async (_graph, flow) => {
    const aOrB = await Promise.race([
        flow.askBoolean('A').then(() => 'A'), 
        flow.askBoolean('B').then(() => 'B')
    ])
    flow.print(aOrB)
})

We correctly get 2 sets of buttons, but no matter which one is clicked, B is always printed (it was created last and therefore is popped first)

As a possibly related issue, clicking a button that was displayed during a previous run will also resolve the ask step from the current run

Lora popup: should not auto-disappear when the cursor leaves if it has focus

#59 introduced a new popup for loras and it's very useful, but if you click into the input fields and let go of the mouse to enter values, it will disappear if the mouse accidentally leaves goes out of the popup's bounds.

Instead it should remain open if I have clicked anywhere into the popup and require clicking outside to close it.

Doesn't work with vid2qr2vid animatediff workflow, fails to parse

Hi, see the attached workflow

vid2qr2vid_v2.json

Better vertical layout by default

"Invalid method encountered' - Error connecting to ComfyUI Server

When connecting to the running ComfyUI-Server in Cushy-Studio I get this error (in the terminal):

Traceback (most recent call last):
File "E:\ComfyUI\python_embeded\lib\site-packages\aiohttp\web_[protocol.py](http://protocol.py/)", line 332, in data_received
messages, upgraded, tail = self._request_parser.feed_data(data)
File "aiohttp_http_parser.pyx", line 551, in aiohttp._http_parser.HttpParser.feed_data
aiohttp.http_exceptions.BadStatusLine: 400, message="Bad status line 'Invalid method encountered'"

(Windows, Version from today)

Write an automated test which pulls a random set of workflows from comfyworkflows.com to test for stability

Every single workflow I try to import fails. This is a recurrent problem in CushyStudio. So far I have never been able to import my own workflows into CushyStudio in the two months that I have toyed with it on and off. This makes CushyStudio mostly useless so it's very important that we get this essential functionality perfect.

https://comfyworkflows.com/

Checkpoint dropdown list doesn't fit with 'dense' layout

If I enable the mobile layout, the dropdown list with all my SD checkpoints feels good, but in the dense layout it seems out of the place, too airy compared to the rest of the UI. I also feel that the list could be a lot longer, imo it's fine to take up the whole available vertical space up to the window bounds.

C:\CushyStudio>npm start

> [email protected] start
> concurrently --kill-others -n vite,electron -c yellow,blue "npm run front:dev" "electron -i src/shell"

[electron]
[electron] [🍪] electronPath: C:\CushyStudio\node_modules\electron\dist
[electron] [🍪] icon path: C:\CushyStudio\src\public\CushyLogo-512.png
[electron] [🍪] failed to patch icon ExitCodeError: Command failed with a non-zero return code (1):
[electron] C:\CushyStudio\node_modules\rcedit\bin\rcedit-x64.exe C:\CushyStudio\node_modules\electron\dist\electron.exe --set-file-version CushyStudio --set-product-version CushyStudio --set-icon C:\CushyStudio\src\public\CushyLogo-512.png
[electron]
[electron] Fatal error: Unable to parse version string for FileVersion
[electron]     at ChildProcess.<anonymous> (C:\CushyStudio\node_modules\@malept\cross-spawn-promise\dist\src\index.js:99:24)
[electron]     at ChildProcess.emit (node:events:514:28)
[electron]     at cp.emit (C:\CushyStudio\node_modules\cross-spawn\lib\enoent.js:34:29)
[electron]     at maybeClose (node:internal/child_process:1091:16)
[electron]     at Socket.<anonymous> (node:internal/child_process:449:11)
[electron]     at Socket.emit (node:events:514:28)
[electron]     at Pipe.<anonymous> (node:net:323:12) {
[electron]   cmd: 'C:\\CushyStudio\\node_modules\\rcedit\\bin\\rcedit-x64.exe',
[electron]   args: [
[electron]     'C:\\CushyStudio\\node_modules\\electron\\dist\\electron.exe',
[electron]     '--set-file-version',
[electron]     'CushyStudio',
[electron]     '--set-product-version',
[electron]     'CushyStudio',
[electron]     '--set-icon',
[electron]     'C:\\CushyStudio\\src\\public\\CushyLogo-512.png'
[electron]   ],
[electron]   stdout: '',
[electron]   stderr: 'Fatal error: Unable to parse version string for FileVersion\r\n',
[electron]   code: 1
[electron] }
[electron] waiting for cushy to start
[vite]
[vite] > [email protected] front:dev
[vite] > vite dev
[vite]
[vite]
[vite]   VITE v4.4.11  ready in 756 ms
[vite]
[vite]   ➜  Local:   http://localhost:8788/
[vite]   ➜  Network: use --host to expose
[electron] waiting for cushy to start
[electron] 200

fix scroll locking

LICENSE ?

got things sorted out ! Thanks comfyanonymous, tekakutli , and Exouxas

Here are my goals with CushyStudio :

Make a cool project that I love to play with.
Have fun with the Community (& grow a community ?).
Share a fun journey with collaborators.
No paywall, no restricted features, no crippling the project for the sake of making money.
Ensure the project remains free and open source.
Make some money in a non agressive way from those who can, to keep working and supporting CushyStudio.
I don't want people to be afraid of using CushyStudio because of some license problem.
I also don't want people reselling CushyStudio.
and I don't want to work for free.

=> I'll probably go with GPL + Contributor License Agreement (CLA)

and while it's free and open-source for all, in the meantime, I'll be able to:

Solicit donations, (though a Patreon or github support)
Sell support, either by contract or by incident.
Sell development services, where people pay you to add features to the software.
Sell a non-GPL version for large companies who want their custom build with custom integrations.
~~WON'T DO: sell a premium version with extra feature.~~

open to your feedback !