For webhooks, click the Notifications button in the sidebar of the dash at the bottom. Then add one
For webhooks, click the Notifications button in the sidebar of the dash at the bottom. Then add one for Stream
FI didn’t get any response from Cloudflare, but it looks like this issue is fixed. At least the video contrast while playing seems fine.
WThis option doesn’t display for me, it says in the documents I need a “Pro” subscription.
WBut I cannot find the language “pro” in any of the Stream options.
Sis there any reason the batch API expiresAt response is set to 2 seconds after generation? Im testing the endpoint and everything is successful in generating the batch api token except for the expiry. I dont see anything on cloudflare docs mentioning this specification.
https://developers.cloudflare.com/images/upload-images/images-batch/
https://developers.cloudflare.com/images/upload-images/images-batch/
Cloudflare Docs
The Images batch API lets you make several requests in sequence while bypassing Cloudflare’s global API rate limits.
PHello friends! We have a customer video that has been stuck in "queued" state for ~2 hrs, is there anything we can do to retry this encoding?
CThis is the channel for #stream, not images. Regardless though those tokens last for an hour, it's UTC time though which may be what you're being confused by
SYea im sorry to ask the question in this channel there just wasnt an #image channel. But thank you for clarifying, ill try the endpoint again. 
THello everyone, i want to ask about download url. Because i am using aws lambda to create backend server-side so that i cannot wait for the time of API response. I have heard about webhook but it is not have any doc about download url notification. Is there any solution for me. Because i really don't know how to creato webhook for download url and then it will send to me a notification.
1There isn't a way natively to get a webhook when a download URL is ready. When you receive the video upload webhook, you could setup some sort of process to poll the video's status until the link is ready or something
SWhen generating tokens for videos, is it required to specify access rules and/or supported regions?
ASince there wasn’t an easy way to play WHEP with Stream’s WebRTC,
I wrote a small Go app that pulls the video and plays it with ffplay.
Might be useful for others:
https://github.com/Azunyan1111/go-webrtc-whep-client
I wrote a small Go app that pulls the video and plays it with ffplay.
Might be useful for others:
https://github.com/Azunyan1111/go-webrtc-whep-client
SAnyone know if we can put html layer inside the studio of realtime meeting? Just like how streamyard does?
CIs there some sort of cap on how far HLS DVR playback can go?
I know m3u8 manifests have the 7200 segment limit, but we're not anywhere close to that. I know it's not time-based because some will go for 4+ hours without issue, while others can start losing the first 30+ minutes of the stream as soon as 70 minutes in. And I don't think it's bytes because I've had issues with some small 480p files much smaller than our normal ones that are fine.
Using OBS + rtmps, but have a couple dozen PC sources so not everyone's encoder settings are the same.
Have also tried pushing some pre-recorded mp4s live via ffmpeg/rtmps and have had similar inconsistencies with those.
We don't use dash and I don't know if it's supposed to support DVR but I threw ?dvrEnabled=true on one and it also only went back to the same spot.
If there's not a retention cap, then it has to be something with the encoding but I haven't been able to figure out the commonalities between the ones that have issues. Any ideas what could cause that?
I know m3u8 manifests have the 7200 segment limit, but we're not anywhere close to that. I know it's not time-based because some will go for 4+ hours without issue, while others can start losing the first 30+ minutes of the stream as soon as 70 minutes in. And I don't think it's bytes because I've had issues with some small 480p files much smaller than our normal ones that are fine.
Using OBS + rtmps, but have a couple dozen PC sources so not everyone's encoder settings are the same.
Have also tried pushing some pre-recorded mp4s live via ffmpeg/rtmps and have had similar inconsistencies with those.
We don't use dash and I don't know if it's supposed to support DVR but I threw ?dvrEnabled=true on one and it also only went back to the same spot.
If there's not a retention cap, then it has to be something with the encoding but I haven't been able to figure out the commonalities between the ones that have issues. Any ideas what could cause that?
1Go to the metrics tab of one of your live inputs and check out the Upload-to-Duration ratio. Would be interesting to know if your client is sending wildly fluctuating packets. I do at least a dozen 3+ hour live streams a week, use dvrEnabled=true and don't have the issues you're describing
1Zoomed in a bit to make the image a bit more helpful
1Can you provide an example of OBS settings that have given you issues?
CA current example, actually. Not my PC nor me streaming. Up for nearly 3 hours at this point but could only go back less than 1.5 hours.
1woah that upload to duration ratio is not great at all
CHas also happened with NVENC H.264 encoder today. Need to check on the rest of their settings. But had the issue with the left picture. No issue on a different stream by same person after that, on the right.
1Are you working with mobile networks or something?
1on my end the lowest ratio I see is like 90+% and can go as high as 150%
1Are these users located far away from you geographically? I wonder if general network routing could cause something like this
1This is from the same stream I sent before now its been running for a while
1oh your keyframe intervals also wildly fluctuate
1I wonder if your users are seeing any dropped frames in OBS
1x264 could totally contribute to an inconsistent encoding output
1especially if the hardware is being strained in any meaningful way while doing the encoding
CTroubleshooting page
"It should be a consistent flat line below 90%. If you see an inconsistent or wavy line, or a line that is consistently above 100%, try the following troubleshooting steps:"
Admittedly they're not flat but they are under 90% most of the time, thought over 100% was potentially problematic.
Not mobile networks. Pretty spread out geographically. No dropped frames listed in OBS ever.
My guess has been it's related to keyframes/open GOP/b frames but haven't been able to pinpoint.
"It should be a consistent flat line below 90%. If you see an inconsistent or wavy line, or a line that is consistently above 100%, try the following troubleshooting steps:"
Admittedly they're not flat but they are under 90% most of the time, thought over 100% was potentially problematic.
Not mobile networks. Pretty spread out geographically. No dropped frames listed in OBS ever.
My guess has been it's related to keyframes/open GOP/b frames but haven't been able to pinpoint.
CMy personal ones never fluctuate with NVENC, which is probably why I've not been able to recreate the issue myself through OBS.
Will have them try keyint, min-keyint, disable scenecut and maybe disable b frames. and see if we can get a constant and closed keyframe and what happens.
Will have them try keyint, min-keyint, disable scenecut and maybe disable b frames. and see if we can get a constant and closed keyframe and what happens.
CAre your keyframes always constant @1984 Ford Laser ?
1Direct from OBS no
1But also working with mobile networks
1
CStill far more stable though.
1yeah absolutely
CCool. Appreciate your time and help!
1No worries mate happy to help
1@claass another thought I had, do you create the live inputs for your users? Live inputs use Durable Objects under the hood, so I wonder if you're creating a really far away DO instance that's causing less than ideal routing for your users. You aren't broadcasting from outback australia by chance?
CWe do, will keep that in mind.
I was able to recreate the issue myself overnight (on a live input I created myself). Switched to x264 on simple mode and got this. Then did a subsequent one on advanced with keyframe = 2 (hence the big keyframe dropoff and leveling) but still had the issue again.
Retrying now with x264 options keyint=60 min-keyint=60 scenecut=0 closed-gop=1.
I'm thinking B frames are crossing GOPs. But I'm not sure that explains the issue happening for NVENC/QSV encoders since to my understanding those are supposed to be closed GOPs by default.
I was able to recreate the issue myself overnight (on a live input I created myself). Switched to x264 on simple mode and got this. Then did a subsequent one on advanced with keyframe = 2 (hence the big keyframe dropoff and leveling) but still had the issue again.
Retrying now with x264 options keyint=60 min-keyint=60 scenecut=0 closed-gop=1.
I'm thinking B frames are crossing GOPs. But I'm not sure that explains the issue happening for NVENC/QSV encoders since to my understanding those are supposed to be closed GOPs by default.
Hcan i use stream as like aws mediaconvert? and also images service to handle tranformas
Hi don't want to deal with ffmpeg and sharp tho
1I think using Stream purely as an encoding service only is against ToS but don't quote me, I swear I read it somewhere
Hhmm need an official answer tho, like docs doesn't mention anything about it
Hthe real thing is im also using r2 too tho so i already store the users data
Hjust need a process for both
