When Civitai asked “How should image creators earn?”, I expected the discussion to focus on payment systems.
Instead, the community kept returning to a more fundamental question:
What is a reward based on visibility worth if good work is not being seen fairly in the first place?
I wanted to understand the discussion as a whole, rather than forming an opinion from whichever comments happened to be near the top of the page. With some technical help, I archived and reviewed every publicly visible comment and reply on Civitai’s proposal article.
At the time of this review, the discussion contained:
316 visible comments
170 top-level comments
146 replies
178 participating accounts
This is not a scientific survey, and it should not be mistaken for one. The participants are self-selected, deleted or hidden comments are unavailable, and reactions are not votes on a formal set of choices. Still, the discussion is large enough to reveal several strong and remarkably consistent concerns.
The original discussion can be read here:
Proposal: How Should Image Creators Earn?
The clearest message: discovery comes first
The most consistent theme was not opposition to rewarding image creators. It was concern that reaction-based compensation would inherit all the weaknesses of the current discovery system.
Users repeatedly described the same cycle:
An image receives a brief opportunity to appear in a feed.
Early visibility produces reactions and followers.
Those reactions create more visibility.
Creators who already have an audience continue to gain momentum.
Good work from smaller or newer creators disappears before many people see it.
If compensation is added at the end of that cycle without addressing its beginning, the platform may reward visibility more than quality.
This concern appeared in discussions about search, Featured Images, follower feeds, recommendations, tags, posting volume, and the short lifespan of new uploads. Many users were not asking Civitai to guarantee equal results. They were asking for a more credible opportunity to be discovered.
My own numbers helped put that problem in perspective
I can only speak directly from my own experience. In roughly six months, I gained 348 followers and received more than 6,600 profile reactions. I am fortunate to receive more engagement than many accounts.
Even so, when I examined my latest 100 images, the median was only seven reactions, and nearly one-quarter received fewer than five. A few images performed very well, which suggests that people do engage when the work reaches them. The difficult part is reaching them consistently.
That does not prove an algorithm is deliberately suppressing anyone. It does illustrate how uneven exposure can be—even for an account that is growing and receiving regular engagement.
If this is difficult for creators who already have some momentum, it is reasonable to wonder what discovery feels like for someone posting their first ten images.
Featured Images are carrying too much weight
The Featured section was one of the most emotionally charged subjects in the discussion.
Several users believe that a relatively small group of creators appears repeatedly while other strong work rarely receives the same opportunity. Whether every perception is accurate or not, the perception itself matters: a discovery system loses trust when users cannot understand why something was selected or how they might receive a fair chance.
Suggestions included:
Limit how frequently the same creator can occupy Featured positions.
Reserve spaces for creators who have never or rarely been featured.
Broaden the range of styles, ratings, formats, and aspect ratios shown.
Add rotating community-curated or personalized discovery sections.
Explain the purpose of Featured more clearly—is it editorial recognition, promotion, discovery, or all three?
Featured should not have to solve discovery by itself. The intensity of the argument may be a symptom of the platform offering too few other dependable paths to an audience.
Following an image creator should mean something
Another practical complaint was that following an image-only creator does not reliably bring that creator’s new work back to the follower.
That weakens one of the most important relationships on any creator platform. A person has already said, “I want to see more from this artist.” The platform should make that choice useful.
A dependable following feed, optional upload notifications, and a clear list of recently active followed creators would help established relationships survive beyond the few hours surrounding an upload. It would also reduce the pressure to post constantly merely to remain visible.
Rewarding reactions sounds simple—until incentives change
Many commenters liked the principle of rewarding images that people genuinely enjoy. The concern was implementation.
Once reactions acquire meaningful monetary value, behavior changes. Bots, alternate accounts, reaction rings, coordinated groups, repetitive posting, and content designed only to exploit the prevailing formula all become more attractive.
This does not mean reactions can never contribute to rewards. It means raw reaction totals probably should not be the complete system.
Ideas raised in the discussion included:
Daily or weekly earning caps.
Diminishing returns for unusually high posting volume.
Weighting sustained engagement differently from a brief reaction spike.
Requiring meaningful account history or Creator Score.
Detecting coordinated reaction networks.
Limiting rewards for near-identical variations.
Rewarding saves, collections, comments, remixes, or prompt use alongside reactions.
Giving older work a longer opportunity to earn instead of closing the window almost immediately.
Several commenters summarized the risk well: rewarding popularity before fixing discovery could strengthen the same feedback loop users are already criticizing.
Quality versus quantity is becoming impossible to ignore
Creators openly acknowledged that feeds can be flooded by numerous variations of the same prompt or concept. This is not necessarily malicious. When attention is scarce, posting more images increases the chance that one survives long enough to be noticed.
The platform’s incentives therefore help create the behavior users dislike.
Possible improvements included daily posting limits, grouped variations, albums, diminishing visibility for repeated concepts, and stronger incentives to curate before uploading. A good solution should avoid punishing legitimate series, model showcases, comics, or experiments while making it less advantageous to fill the feed with minimal variations.
This is difficult, but ignoring it would make any popularity-based reward system much easier to game.
Blue Buzz is widely seen as a reward without enough purpose
A second major cluster of comments concerned Blue Buzz.
Many image creators already receive Blue Buzz but feel restricted in how they can use it. The separation between Blue, Green, and Yellow Buzz is confusing to some users and particularly frustrating to mature-content creators.
Common suggestions included:
Let Blue Buzz purchase cosmetics, profile decorations, and other non-transferable benefits.
Permit Blue Buzz tips.
Allow Blue Buzz to cover part of a purchase through a deliberately unfavorable conversion rate.
Restore broader generation uses, including mature generation, possibly at a higher Blue Buzz cost.
Let challenge hosts offer Blue Buzz prizes.
Use Blue Buzz to promote underused models or support community activities.
Making Blue Buzz useful does not necessarily require turning it directly into cash. Non-transferable utility could reward participation without creating the same incentive for bot farms that a freely convertible currency would create.
A simple thought experiment: one Buzz per reaction
I wondered what one Yellow Buzz for each legitimate image reaction would mean in practice.
At more than 6,600 reactions, my entire six months of activity would still not purchase many of the 9,995-Buzz cosmetics currently in the shop. My median image would have earned seven Buzz, while my strongest image would have earned roughly 99.
That does not make one Buzz per reaction the correct answer. It does show that a carefully limited reward may be more modest than it initially sounds. The hard problem is not the arithmetic; it is determining which engagement is genuine and preventing volume from overwhelming quality.
Many users want the fundamentals fixed first
A substantial portion of the discussion was less interested in a new economic system than in dependable basic functionality.
Commenters mentioned unreliable feeds, upload and publishing problems, notifications, moderation mistakes, rating friction, search quality, weekend instability, payment limitations, and mature-generation restrictions. Their message was straightforward: a new reward system will not restore trust if the everyday experience remains frustrating.
This does not mean Civitai must stop developing anything new until every bug disappears. It does mean that compensation cannot be considered separately from the systems that determine whether creators can post, be found, and maintain an audience.
Not everyone believes image creators should be paid
The disagreement deserves to be represented honestly.
Some users see image generation as a hobby and do not want financial incentives changing its culture. Others believe model creators contribute substantially more labor and should remain the primary recipients of compensation. Some argued that direct tips already provide an adequate mechanism.
There is also concern that monetization would encourage creators to chase formulas, trends, and fetish niches rather than make work they personally value.
These objections are not merely hostility toward image creators. They identify a genuine design challenge: how can Civitai recognize useful creative participation without converting every interaction into a transaction?
Recognition, discovery, cosmetics, achievements, useful Blue Buzz, contests, better analytics, and stronger creator-to-audience connections may matter as much as direct cash for many participants.
Ideas worth exploring
Across the discussion, several proposals stood out because they connect rewards to useful behavior rather than raw volume:
Reward images selected by model creators as strong showcases of their resources.
Share a small portion of model-generation fees with images that successfully introduce users to that model.
Recognize prompt reuse or remixes while preserving attribution.
Reward thoughtful testing and reviews of underused models.
Improve creator analytics so people understand how viewers discover their work.
Create optional subscriptions, paid galleries, marketplaces, or commissions for creators who want them.
Expand contests and themed challenges with transparent rules and controlled prize pools.
Improve tagging through community review and better upload-time suggestions.
Build personalized recommendations that can resurface older work instead of treating images as disposable after one day.
None of these is immune to abuse, but they provide a broader vocabulary than “pay for likes.”
How I summarized the discussion
Most people are not going to read hundreds of comments and nested replies. I normally would not either. Here is the process I used with the help of an AI assistant:
Retrieve every visible page of comments. We used Civitai’s authenticated interface and continued pagination until no new comment IDs appeared.
Preserve the conversation structure. Top-level comments and replies were kept separate so disagreements were not mistaken for independent endorsements.
Remove duplicates and formatting noise. Comments were identified by their Civitai IDs, while HTML formatting was converted into readable text.
Group recurring subjects. Discovery, Featured Images, Blue Buzz, reactions, bots, posting volume, mature generation, site reliability, and alternative reward systems emerged repeatedly.
Read both popular and unpopular positions. Reaction totals helped locate comments that resonated, but they were not treated as votes or allowed to determine the conclusion by themselves.
Look for disagreement and exceptions. A useful summary must include opposition, uncertainty, and minority proposals—not merely the most convenient consensus.
Verify claims against the archived comments. Numerical claims came from the collected data, while interpretations were written as observations rather than scientific findings.
You do not need to be a programmer to use the same general method. Save the complete discussion if possible, ask an AI tool to group recurring ideas, require it to identify disagreements, and then personally read representative comments from every group. AI is useful for organizing a large conversation; it should not be trusted to manufacture consensus or replace the people who participated.
Where the discussion appears to have landed
Civitai asked how image creators should earn. The community’s most consistent answer appears to be:
First give creators a fairer chance to be found. Then design rewards that do not turn visibility problems, spam, and manipulated engagement into an economy.
That does not require abandoning compensation. It suggests an order of operations:
Improve discovery, following feeds, search, and recommendation diversity.
Reduce the advantage of flooding feeds with near-identical work.
Make Blue Buzz genuinely useful without making abuse profitable.
Strengthen bot and coordinated-engagement detection.
Test limited creator rewards with transparent rules and published results.
Adjust the system based on actual behavior rather than assumptions.
Civitai staff stated in the discussion that they are reading the feedback and plan to return with the strongest ideas and a clearer direction. That is encouraging.
The community is not short on ideas. What it appears to want most now is evidence that those ideas have been heard—and that the next system will begin with the people whose work is currently hardest to find.
Snapshot reviewed August 1, 2026. Counts refer only to comments publicly visible through Civitai at the time of collection.