Hi stupid question but where do i access the leader board?
There is a view all stats button on the sidebar and if you go to the stats page of stash and refresh youll see it
The plugin seems to no longer limit the possible selections the filtered performers on the active page like it used to.
Is this a bug or intentional change?
Intentional. Url parsing caused too kany issues with stash itself. Itll be brought back in a new way.
Okay, glad to hear itāll make a comeback. It was one of the best features.
1 Like
thank you Sakota, I installed 1.2.6 and immediately notice how much faster it is, significantly faster so thanks you for all your optimizations
1 Like
Is there a future plan to add images to have the same modes as performers? Thank you
1 Like
sorry if this isnāt the place to ask for help but Iām unable to get scene mode working. I get this error āError: GraphQL Errors: Cannot query field ācustom_fieldsā on type āSceneā.ā
Any assistance to help diagnose this would be great - thank you!
Im really at a crossroads on how or if I want to do images at all. Galleries will most certainly get an implementation but when you have 2mil + images like me doing comparison ratings on images individually seems absolutely insane. When youre talking comics or anything like that it quickly loses its sensibility as well. I think a better option would be to do galleries and leave images alone individually or at least come up with a new approach entirely for images. So, its still on the chopping block.
This seems like an issue with your stash version. The latest version of stash uses custom fields for scenes which the plugin uses. If Ascension cant find this graphql value I imagine it crashes. You should verify your stash version and update or use an earlier version from previous releases for the plugin to work. 1.2.5 scene rating was introduced. If you want the plugin to work without scenes you would have to do a version earlier than that.
1 Like
Oh ok thanks for your reply
ā2mil+ imagesā is insane!
Howās this for an alternative approach to spark some inspiration, Sakoto: a passive rating system?
A system that records the userās active lingering duration on a given image to passively deduce their favorites over time instead of relying on manual comparisons. Yes, this manual āHot or Notā approach is half the fun, but it definitely doesnāt work well for large libraries and the plethora of conflicting image contexts. Thatās why keeping āHot or Notā for performers, and then possibly photo subcategories (e.g., attributes), would be the best approach for a manual comparison mode.
So, the rating could be calculated on a bell curve based on your longest-gazed-at images. That way, it non-intrusively gathers information in the background for large library holders like yourself, then compares those images against others based on viewing duration, providing useful āvalue/rankā data.
Gaze duration, and perhaps even where the user zooms in, could be interesting data points too (thumbnails or image crops?).
It could also help users identify and delete their least-viewed photos that are quickly skipped over.
Whilst writing this, it reminds me of Spotify Wrapped, which captures a lot of information about what you listen to and provides insightful metrics on your most-listened-to songs. Features like that could also be implemented using data like this, which is pretty neat.
1 Like
I couldnāt find documentation on how to select a specific Tier or multiple Tiers for head-to-head mode. I noticed I triggered āTier Selectionā at one point in the logs but could not reproduce it.
How would I do this the correct way?
Before I shoot you down I want you to understand my philosophy when designing the overall direction of the plugin matchmaking\features. You canāt build a house without a blueprint.. well you could, but it may not be what you want.
The 3 key things i consider are the following:
- Discovery: For finding new content, which is important for users with extremely large libraries and preventing bias
- Variety: For giving the user a healthy selection amongst the discovered pool
- Recency: For giving the user fresh selections while avoiding repeats with said variety and discovery
Combining these three factors maintains a stable and healthy system over time and keeps the game from getting stale while promoting engagement. You lose any of these three, and the system falls apart.
I certainly appreciate the thought but at a high level glance over this design seems to fail discovery as the user would have to make a selection thus implying bias as the user already knows what they want. It fails variety and recency alone for the same reason I just mentioned.
Now if you served a queue of images and the āstareā calculation is based off that, for users with large libraries it could become meaningless as they could be skipping for a long time before they find something to āstareā at. The system needs to include the three pillars for it to be a healthy system imo.
Manual Tier selection is not a feature, though I do like this idea. What you saw in the event logs was the tier selection shuffle array moving onto the next tier after the next selection trigger popped after a value tripped. This causes a tier selection to appear in the event logs.
Perhaps I could add this in the next update. But it might be a while. I have a pretty high profile exam due this thursday ive been studying for last few months so I havent spent any time in dev. Just studying. But hopefully if all goes well I can hop back in soon.
2 Likes
Passed the exam with only 4 mins left of the proctored 140mins. Will likely get started on the next update mid next week or the week after.
3 Likes
Have run into what I think might be some sort of bug tracking win/loss rate, unless Iām misunderstanding the system. Added a new performer, and ran a āPlacementā series from the performer page; performer won multiple matches (~5-6ish) and lost one; however, badge on performer page shows only a single match played, with a single win (image attached). Let me know if more information is needed.
Does the performer ledger on the performer profile page reflect the matches?
Iām not sure if this is the āperformer ledger,ā but hereās the Custom Fields output; it doesnāt seem to reflect the matches.
Whats your ascension version from the settings?
Per the Settings screen, 1.2.6.
1 Like
Nice ok thanks for checking ill give it look once i start the next update later this week.
1 Like