This is not complaining:

As many other uploaders, makers, shapers, and other digital degeneracy enthusiasts have noticed we are now on a significantly different trajectory on this site.

On the whole these changes do not materially effect me or what I do on here. I do not interact with yellow buzz beyond what trickles in from my rather sizable upload count. I get free space to host what I make and the traffic what ever my degeneracy brings to the site or helps retain it benefits the site. This is on the whole, a mutually beneficial arrangement.

The net value of content posted to Civitai is by its nature, transitory. New content drives interest, shared via other sources in small doses via X, Discord, or whatever media feed of your choice this ultimately is more practical for most people than actually visiting a website outside of the use and function of it in the first place. This is unfortunately, the reality that faces Civitai.

I know for many who use Civitai, specifically for the onsite generator, it is likely just one of many they visit in a rotation of many. Gathering the daily use points then tinkering with whatever their pool allows them to. Monetizing Civitai is a difficult proposition both from a advertisement and subscription model standpoint.

Most if not all users of this level of content run adblock, have little interest in paying for generations that can be done for relatively free, have little loyalty for hosting content outside of the professionals you do so to make money, and/or spread their content across every possible platform they can to spread their name and recognition beyond what a single niche hosting site can give them.

This leads us in part to the current issue on civitai, where on paper they even to a casual user they could be losing up to 3$ a day from the prior yellow buzz system every daily reset. Given the thousands of users that trickle back from creating or adding content is just too big a burden to shoulder when said users do not generate more revenue even over a longer period than what was being credited to them.

Cost Efficiency and Value Proposition:

All of that said, there is likely a significant means to increase the value proposition and cost efficiency within both the generator, training tool, and content hosting in general. A few things for consideration worth mentioning are often complained about and solvable via segmentation of content, focus of scope in services, and general community participation.

The first consideration I would suggest is not caving to advertisement limitations while also offering different embedding tools for site side advertisement. The use example I point to on this is something i've seen implemented on Royal Road, and a few other sites, where there are three types of ad layers in the embeddings of the site; Paid network ads, affiliation ads, and user internal ads.

Paid network ads are the usual 3rd party advertisements or traditional ads fed through either adsense or otherwise marketed to anyone without an adblock.

Affiliation ads are layered under these advertisements and usually internal redirection or 2nd party style ads for affiliated sites under the same 'network' of the site owner, like if there was redirects to particular parts of civitai as a site, green, specific video tools, announcements etc.

User internal ads are done as internal currency or rewards for reaching milestones and are the bottom most layer of advertisement you can see in place of normal ad's even through adblock much like the affiliation ads.

This can be layered in any metric of preference preferred and would deplete reserves of buzz on the books yellow or blue as desired. I'd suggest doing both with blue buzz of course being either higher cost or lower priority in the ad stack and never showing outside of adblock.

Cost measures

Something else to aid the sheer cost is image compression after the 30 day window on any image with metadata and expanding available tools for recreation of images with shared meta-data. I think it is in at very least in Civitai's interest to encourage new users learning from looking at popular image meta data to expand on the use cases not only for using the site generator but also tagging what tools were used off site if applicable to uploaded content.

So if you have a plugin or extension not only to make 3rd party uploads easier than manually dragging 20 at a time to the post window, but gather more settings data; Kohya methods, ForgeUI internal settings, workflow links, and smaller configuration tweaks potentially including hardware would greatly give the site a greater site wide feeling of showing how you make good things, but also potentially diagnose generation issues on a lora/checkpoint, and increase compatibility with anything posted and hosted on site.

This too leads to a question of cost efficiency in hosting; opt-in programs for dead content removal. If you generate content off site and upload it, is there value in a buzz sink yellow or blue, for preservation of it with no reactions, collections, or other engagement? If you don't sink some buzz, are not a paid membership, or keep up certain metrics, should there be a decay system for your content? either reduced file size, image shrinkage, or some other cost metric reduction while not just removing the content wholesale.

The other big hit that I think we'll see with the change is incentives for botting. With yellow buzz now only being tied to certain additive asset hosting and creation, I think botting is going to mostly die outside of a few nerds testing detection systems or otherwise pushing engagement systems. This is going to be still important for cash prize contests and so forth since even just natural bias of viewing top reactions and the way the filter/sort system works on Civitai can cause issues.

The nice to haves

I think there would be some nice to have options as far as user retention and integration goes, both on the filter system, searching tags, and content promotion side go. Most notably in the vein of meta data tag searching over the official tag pool. This also includes adding functionality for video's to get meta tagging of some kind beyond what user input is given on either process used or tools.

I also think there's monetization potential with a Red section designed to curate specifically around the weird things people want while maintaining the legal limitations of content on site. The most common argument I've seen on staff streams is 'the advertisers' and revenue considerations. Most assuredly its credit processor related most specifically VISA/MC terms of site content that really need to be fought more harshly but it isn't necessarily civitai's burden to fight that battle.

I think some expansion of the buzz bounty system might help, as well as services for blue or yellow buzz including but not limited to custom requests, and trouble shooting services. Offering for up front bug fixing, set-up training, and promptgineering out side the paid service affiliates might be worth considering to add some legitimacy to teaching basic generation tools.

I also think adding some minor perks purchaseable or earned via blue buzz to further push engagement might still be worth while for some people. Even just recognition that 'hey you haven't missed a single day for x days ' badge or hitting certain reward benchmarks could do much to push engagement beyond just 1girl posting and such.

In the End

I think civitai can pull through but it needs to pick up on some of where the wind is blowing in the space and make sure it's not outside of the zeitgeist of what is going to keep them relevant long term as far as a platform goes. Especially with the deepseek visual model testing I've seen going around but yet to see examples of those models either on site or results of said work posted here.