<Shreyas[m]> "Thanks for the feedback! I’ll..." <- Shreyas: aerozol is right about the visibility part. Maybe look for some logic, i.e., given a color A, Logic(A) -> Color B, where Color B is visible on Color A. Now this in itself is very difficult to determine, but I'm sure there might be some study or implementation that tackles this.
jasje[m]: If this proves too much, we can fallback to having shadows or static color.
<jasje[m]> "Shreyas: aerozol is right about..." <- Got it! I’ll look into a way to automatically pick a readable color for any background. If that’s too tricky, I’ll use shadows or fixed colors instead.
Jigen has quit
Jigen joined the channel
GautamShorewala[ joined the channel
GautamShorewala[ uploaded an image: (110KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/matrix.org/AktKGEsSsOamtBouByOEQUrz/WhatsApp%20Image%202024-12-28%20at%2017.21.59_010c7804.jpg >
GautamShorewala[
hi jasje ,
when we have a song selected in a player can we use the image to set the background for player. If its ok then i will start working on it
what are your thoughts on it
GautamShorewala[ uploaded an image: (110KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/matrix.org/zpxwerMwTdhSRyQWSbZoyRoH/WhatsApp%20Image%202024-12-28%20at%2017.21.59_010c7804.jpg >
<zas[m]> "bitmap: postgresql on aretha..." <- not sure what happened but runit kept trying to restart PG while the old process was still running, I restarted the container which helped, but redis is still having issues ("Asynchronous AOF fsync is taking too long (disk is busy?)")
I think we should just move the test.mb containers off aretha
Shreyas[m] uploaded an image: (306KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/matrix.org/YcxrkGiFXuZAkQOlTTfgTvAI/Screenshot_2024_12_28_22_12_23_260_org_listenbrainz_android_debug.jpg >
Shreyas[m] uploaded an image: (297KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/matrix.org/jraujEPPKQBeklXhktfkWTpl/Screenshot_2024_12_28_22_12_02_170_org_listenbrainz_android_debug.jpg >
Shreyas[m]
Shreyas[m]: jasje: How does this (bp_color_primary) color look? I think it works well for both light and dark mode since the title and subtitle in light mode are using lb_purple.
* Hi jasje: How does this (bp\_color\_primary) color look? I think it works well for both light and dark mode since the title and subtitle in light mode are using lb\_purple.
bitmap[m]
<relaxo5> "I Have a MB API question: Is..." <- it shows up in the XML so sounds like a bug in the JSON fmt, can you open a ticket?
re: attribute-ids, they are guaranteed to be stable while the names can change (it rarely happens though)
jasje[m]
<GautamShorewala[> "hi jasje ,..." <- Hi Gautam Shorewala: I think Shreyas: is also working on fixing the video player at the moment. To avoid conflict of work, it would be better if you guys could declare what you're working on respectively.
<Shreyas[m]> "jasje: How does this (bp_color_p..." <- Shreyas: Looks fine, but will need to see its implementation in action first myself
elomatreb[m] has quit
Shreyas[m]
<jasje[m]> "Shreyas: Looks fine, but will..." <- Alright, I’ll open a PR so you can check it out.
<Shreyas[m]> "Alright, I’ll open a PR so you..." <- I would use the blue/purple tints I suggested over orange. Sorry about the trouble, but orange is not the easiest text to read for people with vision issues
aerozol[m]: Orange should be reserved for highlights. If you really want to use orange you could do something like make the play icon turn orange (instead of changing the colour of the text)
<jasje[m]> "Hi Gautam Shorewala: I think..." <- Cool idea, if you are willing to do quite a bit of work to make sure we never have super gross or unreadable colours :) Did you see that we already have a bit of code to calculate album cover colours? e.g. “huesound”
aerozol[m]: The “Music Neighbourhood” feature in particular uses these colours a lot, and makes some quite nice gradients: https://listenbrainz.org/explore/
relaxo[m] joined the channel
relaxo[m]
<bitmap[m]> "it shows up in the XML so sounds..." <- Will do tomorrow, thanks.
Shreyas[m]
<aerozol[m]> "I would use the blue/purple..." <- No problem at all :)
Shreyas[m]: I’ll switch to the blue/purple tints you suggested. Thanks for pointing that out!
* blue/purple tints as you suggested.
GautamShorewala[
<aerozol[m]> "Cool idea, if you are willing to..." <- will check and work on it
Hey mayhem happy holidays! I’ve done a lot of exploring around and while I do foresee myself working on MIR for a good long while I’ve also explored the latest in essentia and I’m going to evaluate their deep learning tempo estimation code. I’m going to compile a dataset of maybe 50-100 tracks from every genre in Spotify (I think I can still do that with their API) and between Spotify and other sites that bend their data I can
at least benchmark the mode against the Spotify data. I’ll publish the results once that’s done and if it’s a pretty strong competitor maybe we use that
Their deep learning tools are implementations of the latest peer reviewed scholastic research so… in effect it should be as good as it gets without any additional investment of effort
mayhem[m]
hi! That sounds very promising -- and I am also glad its not MTG's own work, but the implementation of someone else's paper. (I'm very much not impressed by any of the MTG tech, tbh). I'm super happy you're focusing on this!
