Need SRT listener as well as SRT output update for mimolive as RTMP is a dead (Flash) protocol. WebRTC via mimocall is not yet PRO standard. Need this to use AWS Elemental Medialive / Mediaconnect. SRT input from remote source to studio for live production and SRT output to AWS Mediaconnect for streaming live to millions. Priority should be for SRT output. Thanks.
Venkat
Hi @Venk Thanks for your comments. SRT is on our list. mimoLive 6.0b1 already offers SRT ingest via the new FFmpeg source.
RTMP is very much alive and the most used live video streaming protocol. The only thing it has to do with Flash is that it was designed at the same time by Adobe. Even AWS MediaLive supports RTMP ingest, according to the docs. I don’t have any experience using it, so I can’t say how well it works or not.
No doubt SRT is important, but it is also much more complicated to understand and control by the users.
1 Like
Hi Oliver & Achim
I’ve just been reading up on SRT, to me the first place to apply if for Bonix would be in mimoCall to improve IQ and reduce latency and dropouts/failures. That counts double if mimoCall is being used to daisy chain multiple Macs running miniLive setups, although I expect that those instances in many cases would be on a local network, but I’ve envisaged setups where I do remote vision mixing from an event in another city and take care of the streaming stuff due to lack of technical skills with the people at the event city location.
There’s seems to be different ways to use SRT and as developers you would be able to hide the technicalities of it from users. BirdDog have developed an SRT system for remote medicine where there’s no software engineers on call to sort ports and IP addresses. They talk about it here on the SRT Videos webpage (second row on the left).
TBH I’ve been a bit nervous to use mimoCall for events because it dropped out on me once (a great many years ago now!) and I had to reconnect the stream.
I’ve been a bit nervous to use mimoCall for events because it dropped out on me once (a great many years ago now!) and I had to reconnect the stream.
True,disconnect happens many times. MimoCall is unusable. even with M1 Pro 12 cores, 64GB RAM, gigabit bandwidth!
Hi @Alastair1 Thanks for sharing your thoughts. We’re constantly evaluating SRT support and it becomes more likely the more it is being requested.
mimoCall is based on WebRTC. Latency-wise you can’t get much lower and SRT is no latency champion despite their advertisement of “low latency” as SRT hasn’t been designed for two-way communications. Their “Low Latency” is compared to other methods of broadcast video transmission such as RTMP and K-Band.
As with all network communication, there is a tradeoff between latency, reliability and quality. If you reduce one, the others go down as well. The problem with SRT at that point is, that the decisions are delegated to the user who needs to know the intricacies of the protocol and its design to set the proper values suitable for their use case.
WebRTC (which mimoCall uses) has been designed for two-way communications at very low latency. It is very reliable. Over the years, we’ve had very few stability complaints. What is a problem with WebRTC is that you do not have any control over how it handles bandwidth vs. quality issues and it tends to err on the side of keeping the connection going, which means it increases compression and reduces quality quite fast to react to any network bandwidth fluctuations and is very cautious at increasing bandwidth use again once it encountered a low bandwidth situation.
While it may be possible to hide a lot of the complexity from the user, the people requesting SRT for mimoLive want it mostly because it gives them more control over the connection paramaters.
We’ve recently updated the mimoCall client and you should see a notable improvement in reliablity and quality.
1 Like
Hi @Venky I can’t confirm the claim that mimoCall is unusable, especially on an M1 Pro. Can you be more specific on how you use it, what client you’re using and what the disconnect looks like?
Here is another request for SRT output. Very important step!
Thanks for the detailed response, Oliver. I just saw it when I was searching my emails for an email with Achim. I take your points about latency vs WebRTC and the engineering tradeoffs that are always there when you have three interdependent needs.
In this case I would rank the three as security of connection (extremely low risk of dropout for 2+ hours) being most important, due to repetitional damage for mimoLive users and for their clients, with latency probably nudging image quality out of the way for second placing. Of course sometimes API/protocol dependencies/choices dictate some of the tradeoffs to certain extent, we’ll take any advantages we can get in an A/B choice of protocols etc, so some tradeoffs can’t be negotiated by the software developer. Take it or leave it :-).
Oliver, I’ll definitely check out and test the reliability of the new mimoCall code. It’s a case of once bitten and twice shy for me, I’m easily traumatised by live events almost going pear shaped and having to improvise workarounds in the moment, leaves me stressed for days and scare tissue! I’m sure others can relate.
On one occasion I held my iphone to a wireless audience mic that was broadcast to the auditorium’s PA system just to keep the session going (it was when the WiFi in the theatre dropped out for a second and cause mimoCall to drop out, this was like in ~2017? and I wasn’t able to reconnect the mimoCall immediately even though the WiFi came straight back up. Fortunately the auditorium wasn’t underground… some I used to broadcast from are underground… so was a cell reception wasn’t a problem and I already had a backchannel open on the phone in case of any mishaps!)
I’m not doing live events these days but plan to find an excuse to start up again when I upgrade my main Mac to an M3.
Would Bonix be able to prerecorded video (or do it live while hosting a seperate test rig) where you stress test an M3 Pro MBP (say for middle price range) it against an M2 or M1… some of the graphics stress testing in CineBench rendering etc is surprisingly good. not sure if the GPU and RAM/SSD speed gains will translate to Multiple In Multiple Out live video work…
I’d love to know where the thresholds are for video broadcasting in mimoLive vs various Macs from budget prosumer to quality prosumer to expensive professional Macs and added on hardware.
Like a matrix of hardware along the horizontal axis and software load on the vertical axis. So on the vertical axis there would be a variation in the number of video cameras connected, live on screen, video FX and other post like QC-based animation/overlays, mimoCalls connected to the session, mimoCalls on-screen, recording quality, ISO recordings, stream quality, (perhaps the number of virtual webcams if that is a use case people have with multiple webcams, I don’t ever do that)
I’m going to make a note to draw up this matrix and at least mimoLive forum users can start posting their own results. like a kind of Geekbench for mimoLive. minoBench if you like!
“unusable” is kind of vague, it’s opinion based not evidence based.
Risk and uncertainty are different things. Risk can be quantified, uncertainty cannot be quantified, just observed as a fact and felt as a subjective reaction/intuition.
We could quantify the risk of mimoCall dropping out on your high end system and network connection(s) by running 1,000 hour long test broadcasts and then counting the frequency of dropouts.
Translating that to other Macs and other network connections is uncertain due to the nature of IP packet transfer under various protocols.