Multiple Source Formats on ATEM 1 M/E production switcher

I need to verify a pre-sales question on the Blackmagic ATEM 1 M/E Production Switcher 4K.

I will have a variety of sources coming in, some 4K over SDI and a few HD (1920x1080) signals over HDMI. Final output will be 1080p, does the switcher act as a scaler? I know that my final output is only one size, but can I have multiple types of sources coming in?

Basically can I set each input to a different source size, but have one final output that are all the same.

Thanks for the help!

Nick

[Gary Adams]

Hello Nick. All ATEM switchers require each input to be the same resolution and frame rate as the selected output. While there are frame synchronizers on each input, there is no scaling capability.

Regards, Gary

Gary, thanks for the response. Just for clarity. I can use either SDI or HDMI inputs at the same time, as long as the resolution is the same correct...

So for instance hook up a camera via SDI but then hook up a computer via HDMI, as long as they are both sending the same resolution we would be good, correct?

Thanks again.

Nick

Gary, thanks for the response. Just for clarity. I can use either SDI or HDMI inputs at the same time, as long as the resolution is the same correct...

So for instance hook up a camera via SDI but then hook up a computer via HDMI, as long as they are both sending the same resolution we would be good, correct?

Thanks again.

Nick

You can use both SDI and HDMI at the same time as long as they're not on the same input as you can only choose one at the time in settings. Also don't forget about "same frame rate" part, that has been mentioned before. It's very important. Most of modern PCs have enough flexibility to match broadcast standards.


Sent from my iPhone using Tapatalk

Awesome, thanks for the help.

Nick

I agree with his emphasizing the frame rate comment. And remember that 60 fps and 59.94 fps are not the same. And your ATEM knows it.

I never understood, and I think I'll never understand in future the resticted function of ATEMs frame synchronizers.

For my oppinion, I should be no problem to support different framerates at the same (!!) resolution .

What is the difference between an unstable source and a different framerate during synchonization ? The source writes into the buffer using its timing and is read using ATEMs internal timing. Normally, for this is used a ring buffer.

Why it isn't possible ?

A frame-sync and standards conversion (or Cross conversion I guess in an HD world) are fairly different processes, and I expect would require significant hardware changes to the ATEMs, and cost increases.

What is the difference between an unstable source and a different framerate during synchonization ?

Data rates & timing & buffer sizes & interlace field order & colour space conversion & proibably other things too.

The source writes into the buffer using its timing and is read using ATEMs internal timing. Normally, for this is used a ring buffer.

Would result in dropped frames or doubled up frames and weird motion artifacts or significant latency on inputs.

The solution you describe is however available. It's called a Graphics Renderer with live inputs, so if you put a Decklink Quad card in a CasparCG along with another channel for output, you'll have no problems cutting up mixed format sources. A Tricaster will handle it too. VMix will do it, PlayBox have some "renderers pretending to mixers" and there are so many others..

A mixer mixes sources line by line, not with a frame buffer, and hence in a fully gen-locked environment, you have what amounts to be zero delay through the system, which you can NEVER get with a Renderer type solution.

As is so often true, it's a case of "right tools for the job".

Why it isn't possible ?

It is possible, it is just undesirable.

Data rates & timing & buffer sizes & interlace field order & colour space conversion & proibably other things too.

My thoughts are only about timing (!), no other conversions (No size, no color space, no interlacing, no ...)

The maximum difference between input and output is one frame, so the frame must have a capacity of one frame. So the buffer size cn't be a problem (at at some older firmware Versions it also wasn't !)

Would result in dropped frames or doubled up frames

This problem also exists when the timing of an unstable source is nearly 0 to the internal timing (so it alternates beweeen before and behind internal sync). But you are right, in my described case it may happen, in case of framerate conversion it must happen.

But maybe the customers could live with it.

So you are simply talking about mixing 1080i59.94 and 1080p30/60.

BMD are on record saying this problem was a bug in the early firmware, they "restored the bug" for original model ATEMs on user request, but it was fixed again in the 4K Range. The bug only didn't cause major issues because 59.94 and 60 are so close, and the ATEMs output were only ever running at 59.94, and hence the computers were always quicker.

You are wrong about the timing between 50Hz and 59.94Hz sources, it's more than a frame, one 50Hz frame is 1.2 times as long as as a 59.94Hz frame, so covers parts of 3 frames. And is why if you put a none sync source into a mixer that doesn't completely reject the input, or has an option to ignore none-sync errors, you'll get a picture that either migrates gradually up or down the output, or flickers like crazy.

As I mentioned you are describing a problem there is already a perfectly good solution for. Graphics Renderers are using the most recent full frame buffer from a source as a texture on a surface rendered to the output context which is timed according to the GPUs output (extern ref with a Quadro SDI or various Matrox cards). A mixer isn't doing that at all!!! It isn't filling up a buffer and then painting that buffer to the output, it's mixing the sources line by line.

The impression your post and many others on this issue gives me, is you are suggesting extreme solutions to the never ending problem getting computer sources into your mixers. IF ONLY BMD would note Tom Bassford's long standing request to use the EDID features of the HDMI on their devices more intelligently, these topics would go away and never come back.

I totally agree. that it wouldn't look fine ! And yes, EDID would be great (for HDMI, what about SDI ?)

But I continue in asking, why it is blocked..... For the ringbuffer itself, the timing from the incoming frame is totally irrelevant.

But maybe it is easier to explain a black input for non matching framerates, than a flickering one...

Exactly how BMD implement the hardwares buffers internally we will never know.

As for the rest of your post, to me it doesn't make any sense.

Perhaps you could explain what the use-case is, that your solution is trying to solve???

But I'll try one last time:

Video Mixers require matching video source formats because of the nature of how line by line mixing works and the requirement to minimize latency through the system.

Computer based renderers use frame buffers to be able paint a new output picture.

There are of course hybrid solutions, where functionally you insert a Teranex processor in front of every input to a mixer and hence use sources of every format possible and by setting the Teranex output format to be the same as the video mixers format, you can mix them all up. Each input would have a processing delay depending on the algorithm required to go from Source format to Mix format.

There are mixers available that have this pre-process stage built in, at least on some inputs, but this adds to the cost & complexity of the device, so why would we want BMD to do that???

I always rant on these topics. Sorry about that, but the ATEM functions 100% correctly with respect to requiring matching video modes, and suggesting there exists tweaks BMD could make to the video processing side, is wrong and misleading. If they are going to change anything to make getting PC outputs into the ATEM easier, it should be the EDID strategy, both on the ATEMs and the mini-converters (to answer your SDI point).

Thank you for the exlaination how, live video mixing works. Maybe it is caused in my bad english, that you don't understand my use case and the resulting question.

I live in PAL-land, so I don't have any problems with 60 vs 59.xx fps. But a lot of people here do have....

So I'll try it also one last time.

The framebuffer can synchronize a timing mismatch up to one frame. Why can't the same framebuffer without any modification (expect removing the format checking) then solve the 60 vs. 59.xx situation ? It is not more than correcting a mismatch of some percents. When I remember correctly, it worked already (in some models ?)... But was removed like a bug.

When this could work, a quick and dirty (!!!) conversion between 50 and 60 formats could also be possible.

Why is this undesirable ?

What is the use case ? ATEM models are very interesting for low budget productions e.g. in schools, sportsclubs,... because of their low price. They don't do broadcasting television, they may want to do a simple livestream with some borrowed stuff and have problems to connect them.

Yes you are right, the right way is to use exactly matching equipment - optimal, if genlocked. For connecting non matching formats, there are convertes and scalers.

So the only answer to my question is your "Exactly how BMD implement the hardware buffers internally we will never know.

The framebuffer can synchronize a timing mismatch up to one frame. Why can't the same framebuffer without any modification (expect removing the format checking) then solve the 60 vs. 59.xx situation ? It is not more than correcting a mismatch of some percents. When I remember correctly, it worked already (in some models ?)... But was removed like a bug.

It was a bug!!!!!!!!! It only worked because the difference in field length between 60Hz and 59.94Hz is 0.2 microseconds, the difference between 50Hz and 60Hz, is 3.5 milliseconds. The difference is two orders of magnitude.

When this could work, a quick and dirty (!!!) conversion between 50 and 60 formats could also be possible.

Why is this undesirable

You continue to assert that that it's possible to "Mix" 50Hz and 60Hz video when it is not. There is no "quick and dirty" conversion. The undesirable thing is to add a Teranex style process to every input, and hence the "cost and complexity" I previously mentioned.

What is the use case ? ATEM models are very interesting for low budget productions e.g. in schools, sportsclubs,... because of their low price. They don't do broadcasting television, they may want to do a simple livestream with some borrowed stuff and have problems to connect them.

Wrong!!!!!! ATEMs are used in broadcast television, as well as online streaming, education purposes, industrial applications etc.. etc.. but that makes no difference anyway. Vision mixing for television or 50 people watching on YouTube or in your cafeteria on your internal loop doesn't change the technology.

And that is not a use-case, a use-case is a scenario or setup or sequence that defines a problem, for example: "I need to produce a show where my cameras and output format are 1080i50Hz and my computer sources are 1080p60"

Solution: Get a Tricaster.

Tricaster cannot handle multiple frame rate sources - all sources must match the operating framerate of the switcher. Tricaster can cope with different frame sizes and aspect ratios but it cannot framerate convert

Framerate conversion is a very expensive effect (in terms of video processing) - you can see this easily using any desktop NLE - load in a clip to the timeline at a matching frame rate and render it to the same framerate - now change the output settings so you are changing the framerate and render it again - you'll notice that the render with the framerate change takes a lot longer to process.

John is completely correct there is no need for vision mixers to handle multiple framerates, one should simply setup all your sources correctly to a common framerate. The only possible exception to this is in corportate event type scenarios where you may have to handle client laptops or video conference systems which cannot be changed to match your operating format. In these kind of environments then you use an event video processing system such as Barco E2 or Christie Spyder to deal with a multiformat environment.

That info on the TriCaster is not entirely true, you can mix some frame rates sources together, for example 23.976fps, 29.97fps and 59.94fps. Mixing different video standards together like NTSC and PAL is not supported.

While I agree that preferable to have all sources operate at the same frame rate, it is a nice feature when you do have the ability to mix when you have some equipment that will not match that other gear.

Kane Peterson
NewTek

Funny part is that original ATEM switchers from ECHOLAB , can mix different resolutions on inputs .. And can individualy turn on/off framesynchronizers .. etc ..

Yeah and had supersource on the 1me. They were much better than the blackmagic switchers, but did cost $20,000.


Sent from my iPhone using Tapatalk

I found that a Decimator MD-HX Cross solved 99% of computer input issues. Not only will it scale HDMI input to an SDI to match camera inputs, but also scales or loops the HDMI input to an HDMI output at the same time, eliminating the need for an separate HDMI splitter. Solved so many issues for me. Hope it helps.