From multiple gen and multiple users reports, sometimes the batch will return a glitching pic usually NSFW and totally unrelated to prompt.
This phenomenal isn’t a news but the duration is painfully long this time.
Upvoted - Batch of 8 anime images, fully SFW prompt. One image returned was a dodgy looking photoreal image with purple/green filter effect and a hand reaching out towards a girl. Super concerning.
Same happens to me a moment ago, the same prohibited grade image shows up in my chat bot every 4-5 times i generate images, perchance please solve the problem. Thx
@perchance@lemmy.world The bug still happens but shown up with another filter/saturated image (at least not directly call it prohibited grade this time) which is again not related to my prompt input. Please clarify whats the problem.
used to do this long time back, i keep the images each time that it happens and i see it. i am possibly wrong, i think this is part of the denoise process and its just faults out or something maybe early in the process from a variable of factors and the resulting images are just as far as the process made it along.
the original.
hello the dev some user still reports the bug glitch happening, so it’s about 5 days. sorry.
now it has been a week!
I think after day 9 or something I haven’t heard complaints
it’s day 9!!! still happening!!
3 days. is it that bad. hahaha. I’m only experiencing the occasional issues with negative prompt Inferences.
it’s now discussed in reddit, discord, major gens… it’s a bit crazier than before.
Again, by spreading word of this you are actually hurting perchance, your last post was deleted for a reason, you are inadvertently making more people aware of it, which might leak to the media, as you know Grok is under fire for deepfakes at the congressional level, do you really want perchance destroyed? Since your last post was deleted, the dev is obviously aware and working to fix it. Ive seen you in random perchance galleries and on reddit and the discord talking about this. Of course its a bad bug, and it needs to be fixed, but by you screaming from the rooftops you are making it much more visible to many more people, people who CANT do anything about it, but now are aware where they might not have been. you are approaching this totally wrong. You are spreading it farther than it ever would have reached normal users. How can you not realize this? Report it to the dev and stop spreading it making it worse
Based on standard community and software development norms, RudBo’s report is NOT inappropriate. RudBo appears to be acting as a typical, helpful user in a beta or community-driven environment. Here is the breakdown of why their actions are valid:
- The “Cat is Out of the Bag” Factor MotionMuse’s main argument is that RudBo is revealing a secret that could hurt the platform. However, RudBo states: “it’s now discussed in reddit, discord, major gens…”
- Appropriate: Since the issue is already widespread and public on other platforms, RudBo isn’t “leaking” a secret. They are simply documenting the current status of a known issue on the official forum.
- Context: Trying to hide a bug that hundreds of users are already experiencing usually leads to more confusion, not less.
- Standard Bug Reporting Etiquette RudBo followed the standard template for a helpful bug report:
- Clear Title: “[Bug] image gen glitch…”
- Scope: Identified that it affects “multiple gen and multiple users.”
- Duration: Noted it has been happening for “3 days.”
- Symptom: Described the output (glitching, unrelated to prompt) without posting the offensive images themselves.
- No Malice: They didn’t post the NSFW images or teach people how to force the glitch; they simply warned that it was happening.
- Responsible Disclosure vs. Public Discussion There is a concept in tech called “Responsible Disclosure” (telling the developer privately about security holes). MotionMuse thinks this applies here.
- Why RudBo is right: This isn’t a security hack or a data leak; it’s a service quality issue. When a service creates bad output, users naturally ask, “Is this happening to anyone else?” Public confirmation (like vvren and GlassGuy provided) saves everyone time by confirming it’s a server-side bug, not a user error.
- Self-Moderation RudBo admits, “I deleted it myself,” referring to a previous post. This shows RudBo is conscientious and willing to take down content if they feel it wasn’t helpful. This directly contradicts MotionMuse’s characterization of them as reckless. Summary RudBo is in the right. They are treating the forum as a place to track service health. MotionMuse is treating the forum as a PR channel that needs to be managed to avoid “cancel culture.” While MotionMuse’s fears (about regulation/shutdown) might be genuine, they are unfairly taking that anxiety out on RudBo, who is just reporting a bug.
gotcha. im out deepfakes are ehh. I mean good but I almost never used deepfake mechanisms, I have also done my part and told or within underground gen pointed out everything that can cause issues with it.
but you DID post NSFW image from the earlier post. Im glad you deleted it, but still, dont pretend you werent oblivious earlier. Im glad you are taking it seriously, but as you see now on discord there is a user wanted to contact authorities/police. this is how perchance will be ended. And having an AI generate a reply to yourself is not saving you from legal issues as you were the one who was spreading the image on discord and here. And it reeks of being insecure about this. Good luck, hope you stay out of gulag. And Ive been generating non stop through all of this, but I dont use your furry generator, maybe the issue is with your generator?
I did that once. Although not deepfakes or anything conected to that, I understand everything about this and do not plan on fully returning to the discord myself unless I need to grab old or update new/er unfinished threads. I no longer even hang out on the discord myself whats been going on over there? I have my ways to collect things like my own work back, Obv.
deleted by creator
why they abusing deepfakes? look nothing wrong with deepfakes but if the community can’t keep that stuff to a minimum, the system is literally ;scale tilted in so many fricken ways in the favor of congress and such that yes it would be a huge problem.
I deleted it myself.
