I am sorry if this too vague. But unlike stable diffusion the open source language model community doesn't have a gifted website such as the civitai. Where people can upload their finetune models, Lora's, and public can rate them. It would be awesome if the developers could make something like Civitai LLM which is dedicated for open source LLM . As of now people depend on hugging face, which doesn't really give much idea about which is best good, how is it good etc. may be not soon but just a thought I really wanted to share.

Planned

Civitai not just for Stable diffusion

Support Text-to-Video and Text-to-Audio models.

Support Text-to-Video and Text-to-Audio models.

On 14th august the size of the images on feed in the on-site generator were changed to be much larger.
I don't think they need to be larger at the expense of making the generator even worse to navigate.
Please revert this change or make it toggle-able.

Implemented

Generator feed images too large

When I browse images, I want to be able to enter a query in the search bar like " #cute #cat " and view images that have these tags

Search images by tags

Right now, we have "delete" and "post". I'd like to have a "download" button there as well. Thanks

Batch download from image generator

It's frustrating when you look for some cool images to replicate and it turns out that they were generated with other programs like Photoshop or other GUIs. With this filter those posts with no generation data would be hidden automatically.

Filter to exclude post/images with no generation data

Search results on the web seem to indicate that stable diffusion comes with some inpainting and picture2picture features. I enjoy learning creation on your website and wont be purchasing nvidia card and downloading stable diffusion any time soon. Could you please add some inpainting and/or pic2pic option on the web interface? If there are some specific reasons why you can not do that, could you let us know the resons?

Inpainting and Pic2Pic

Please, add a way to communicate with users. Ty

DMs / Friends and Private messages.

Things like this should not be happening. This is WAY beyond trolling.

This is both defamatory and abusive and I'm sure is meant as a form of attack!

IMHO the tags should have a hover feature that link back to the users profile that made it. Make people accountable and make this sort of abuse trackable on the site.

Please fix this:

https://civitai.com/images/506112?modelVersionId=46846&prioritizedUserIds=63217&period=AllTime&sort=Most+Reactions&limit=20

Abusive Tagging

the ability to organize models into collection instead of just having them all in as liked.

Collections

I want to share many models (for example children, clothing or art styles) but I am concerned that some people may misuse them to create offensive images, which I do not want to be associated with. This sometimes makes me hesitant to share my model links with family members or business partners.

Anyone can download any model and do whatever they want but don't post it on the creator model page without permission. I believe that creators should have control over what types of images are associated with their models. 

One solution could be to allow creators to remove images that they do not approve of from their model page. However, this could create other issues.

Instead, I suggest implementing a feature that allows creators to choose which types of images can and cannot be posted on their model page, such as NSFW or copyrighted content. This would give creators greater control over their work and help ensure that their models are being used in a respectful and appropriate manner.

Thank you for considering my suggestion.

Model Category Control: Let the model creator choose what category of image can be posted on their model

I am not mean to attack sb.

But look at the "Guardians Leaderboard", the first one has "1.5k+" shots while "800+" misses, which is about 7 times of the second, and that means lots of models were reported wrongly. 

We all know that it is important to avoid child porn but isn't someone too sensitive? Here, https://civitai.canny.io/feature-requests/p/report-users-warningban-system，two images are nothing to do with child porn but still get reported. 

Report abusing will not only make moderaters tired and feel their time wasted, but also hurt creators' passion and enthusiasm. Creators should not be worried about things like "is the character in my image looks mature enough?" "I should add huge breasts so that it won't be reported and get unpublished".

Need a way to Avoid report system being abused

It would be awsome to be able to turn off the different mature content blur.I dont mind seeing clothed boobs, or even fully naked woman and explicit nudity, but it would be nice, if i could select in which range i wanted to see it, so instead of just blurring all mature content, it would be nice to being able to select which range of nudity i would like to blur, and not

Browser mode. Selecting to blur, 13+, 17+ & 18+

I love browsing the website for new models to discover, but:

- When I sort by "most commented/reactions," it mostly shows popular models that I already know

- When I sort by "newest," the quality is uncertain because no one has liked or commented yet.

I would like to suggest new options: sorting by most comment/reactions but only showing models that I haven't liked or commented on yet. To help with this, I propose adding a "Quick Hide/Unhide" feature to the main page. This would allow users to quickly hide models that they have already liked, making it easier to search for new models that they haven't seen or liked before. This would greatly improve the browsing experience on your website and help users like myself discover new models more efficiently.

Thank you for considering my suggestion. I look forward to hearing your thoughts on this feature.

Awaiting Review

FRESH FIND: Only show models I haven't seen or liked yet

Page scrolling too long can cause browser errors, it is recommended to add pagination mode

browsing experience - Pagination

Crossreferences for wildcards are hardly possible, and I think for workflows, too (when not included in the meta data). Maybe there are some other types of resources, which face the same. So it would be nice, if you could add resources manually in a post, e.g. wildcards.

Adding Resources manually

Basically, as landscape Images generally get the short end of the stick add some side toggles on the left so you have: - 1. All. Show Images as they are displayed currently. 2. Port: This would display Portrait and square format images only. 3. Land. This would display landscape-orientated images only in a 16:9 frame. Mockup example: - <img src="https://fb-usercontent.fra1.cdn.digitaloceanspaces.com/01917c6e-5f1d-7438-b374-3a28c7f54d5f.png" data-width="100%" data-align="left"> I think it would be beneficial for the community as a whole.

Toggle between All, Portrait and Landscape images

In consultation with the good folk of Office Hours, we’re proposing an Advanced or “Unrestricted” Generation mode toggle, available from the profile picture menu &gt; Account Settings. When enabled, the following would be possible;<ul><li>Overriding the Creator recommended Strength for LoRAs</li><li>Enabling the use of models from different (potentially incompatible) ecosystems (e.g. SDXL models + Pony Resources)</li><li>Allowing lower than normal Step counts (sub 10), and other potentially game-breaking settings</li></ul>There would be a high risk of image generation failures, which would be acknowledged by enabling the Advanced Mode.Please feel free to comment your thoughts below, and add your suggestions for other variables which might be “unlocked” in this mode.

Image Generator Advanced Mode

The creator is already plugged into the user's preferences so it can not show ignored models. Should be a relatively simple process to add a one button on/off switch to show only models a user has favorited that support on-site generation.

Option for "filter to favorite models" under Add Additional Resources for on-site generator

GGUF is rapidly becoming the most popular and most performance-friendly quantization method for Flux checkpoints, with NF4 a close second (or maybe I have those reverse, idk…). Regardless, a way to sort by quantization method is required now that we have a popular diffusion model (Flux) that has a wide array of different quants that all perform differently. Please give us a way to sort by quant.

GGUF and NF4 Tag for GGUF models

Why can one not show up his posts in their profile instead of first 4 last images?This would give a much better overview of what a creator is doing. Showcases also only show some pictures. I still do not understand, what showcases are for. Maybe that’s my fault them.The advantage of the posts overview is, that the viewer can click on it and see all other images of that series and react on them.

Showing Posts instead of Showcases or Images in profile.

i.e:instead of"LoRA (36.11 MB) Download"this text:"LoRA (36.11 MB) Download Dragonsv1.safetensors"

Please show the native Filename for the download-links:

There are instances where reports are based on ToS violations, but the system does not allow comments in that context (eg. necessary explanation why this actually is a violation). This would be helpful.

Allow additional information on ToS reports

automatically exclude reviews that do not include text or images

Automated review exclusion

Help Center

Improved

Fixed

Admin

Manager

Contributor

civitai

In Progress

Rejected

Send us your Feature Requests! Please search for existing feature requests before creating a new report, and be aware that many issues on <a target="_blank" rel="noopener noreferrer nofollow" class="text-indigo-500 dark:text-indigo-300 cursor-pointer link" href="http://Civitai.com">Civitai.com</a> are transitory/temporary! If you'd like to submit a Bug Report, please head over to <a target="_blank" rel="noopener noreferrer nofollow" class="text-indigo-500 dark:text-indigo-300 cursor-pointer link" href="https://civitai.com/bugs">Civitai.com/bugs </a>and submit a new support ticket. Note: Bugs submitted here will not receive an answer!