I disagree with this. Text is regularly very blurry and pretty difficult to read, and sometimes that’s luckily fixed when an I-frame arrives (relying on luck and timeouts sucks but it is OK in practice). And worse: jumping mouse cursors which make tracking movements very difficult.
There’s definitely room for improvement. But does it require WHIP? I get the impression the difference would be that the encoder would be smarter and wouldn’t try to use an encoder tweaked for scenes from a webcam with a single person mostly to show presence, when it needs to show sharp screen content. But like, isn’t that just a basic client issue/limitation? Of course two different encoders could be each tuned more specifically but there’s no reason it requires independant processes and a token granting dance.
(but WHIP itself looks nice, especially when you want to introduce another capture since capturing and encoding your screen playing an existing capture is pretty shitty)
Discord has different encoder profiles explicitly with the regular encoder tuned for games/videos and one oriented towards text fidelity (at a lower framerate though, so you do get cursor tracking problems). Zoom seems to default to a text-oriented one, with I think an option tuned towards video.
Interesting. These are the ones I have less experience with. Google meets and Teams have always been pretty bad for screensharing IME. For discord, it makes sense that, by default, it is so tuned towards games and videos that it also needs to detect static content to not be too bad with them.
I disagree with this. Text is regularly very blurry and pretty difficult to read, and sometimes that’s luckily fixed when an I-frame arrives (relying on luck and timeouts sucks but it is OK in practice). And worse: jumping mouse cursors which make tracking movements very difficult.
There’s definitely room for improvement. But does it require WHIP? I get the impression the difference would be that the encoder would be smarter and wouldn’t try to use an encoder tweaked for scenes from a webcam with a single person mostly to show presence, when it needs to show sharp screen content. But like, isn’t that just a basic client issue/limitation? Of course two different encoders could be each tuned more specifically but there’s no reason it requires independant processes and a token granting dance.
(but WHIP itself looks nice, especially when you want to introduce another capture since capturing and encoding your screen playing an existing capture is pretty shitty)
score 1 · canonical
Discord has different encoder profiles explicitly with the regular encoder tuned for games/videos and one oriented towards text fidelity (at a lower framerate though, so you do get cursor tracking problems). Zoom seems to default to a text-oriented one, with I think an option tuned towards video.
score 1 · canonical
Interesting. These are the ones I have less experience with. Google meets and Teams have always been pretty bad for screensharing IME. For discord, it makes sense that, by default, it is so tuned towards games and videos that it also needs to detect static content to not be too bad with them.
score 1 · canonical