Blackmagic Ursa Mini 4.6K - Audio Tests

Thanks to the awesome guys at Empire Rentals, we've been playing with the Ursa Mini 4.6K and the brand new Public Beta 4.0 firmware yesterday and today, testing things out before we press the "buy" button. Our team did a bunch of camera/lighting tests overnight, but the main thing I personally wanted to really focus on is the audio performance. First up, I have to say I'm very impressed. It's LIGHTYEARS better than the Blackmagic Cinema Camera, Blackmagic Production 4K and Pocket. It's no Alexa - but that's to be expected considering the price difference!

Here's some quick observations:

a) The Ursa Mini doesn't have any of the frequency response issues that the original BMCC and BMP4K cameras do, which is awesome.

b) The Ursa Mini is really good for single operator jobs - as you can hook up microphones directly to the camera, and adjust levels on the fly.

c) There does seem to be some phasing issues, but they're easily fixed in post.

d) You can send timecode to the camera, which is awesome - but there does seem to be a delay between picture and audio, so you'll need to do an off-set in post. The annoying part though is that it's samples off - not frames, so you can't just have a frame off-set.

e) The headphone output is much better than previous Blackmagic Cameras in terms of noise level, however it's still not perfect. There's also a massive audio delay on the headphone output, which makes monitoring off the camera almost impossible. Personally, we'll just always hook up an external mixer to the camera, and monitor off that, and only switch over to the camera's headphone output for confidence checking, and calibration. If the camera operator had headphones in, they'd go insane.

Here's a bunch of unscientific tests we did in the office (i.e. not in a sound proofed studio or anything fancy):

1) Our first test was to play our previously made "test" track on our Sound Devices 552, connected via XLR and record it in the camera. This is basically just a quick and dirty test to see how the frequency response of the camera performs. The camera was in RAW mode, the audio input was set to XLR, both inputs were set to Line with a 86% gain. Essentially we just save our "test" track on the 552, press play on the 552, then record on the camera.

You can download the files here:

- Test Track
- WAV file straight from the camera

Here's a screenshot of the test track:



Here's a screenshot of what came off the camera:



As you can see - they're pretty close, which is much better than our previous BMCC tests!

2) The second test we did we again had the 552 connected to the Ursa Mini via XLR. On the left channel we had a Sanken CS3e and on the right channel we had a Sennheiser ME66. Again, the camera was in RAW mode, the audio input was set to XLR, both inputs were set to Line with a 86% gain. We also recorded to the 552 for comparison. Apologies in advance - I'm certainly not a voice over artist!

You can download the files here:

- WAV file straight from 552
- WAV file straight from the camera

3) The third test we did was just plugging a Sanken CS3e and ME66 directly to the camera via XLR. We had the audio input set to XLR, with both inputs set to Mic, both channels set to High with Phantom Power on, the Channel 1 gain set to 86%, and Channel 2 gain set to 96%.

You can download the file here:

- WAV file straight from camera

4) We then did a similar test with a Sanken CS3e and a COS11 wireless lapel via a Sennheiser G3 kit hooked up directly to the camera. This time I fired myself as the voice over artist, and got Nick to take over because he sounds so much nicer than me.

You can download the file here:

- WAV file straight from camera

5) Finally, we did a Timecode test. We had a Tentacle on the Ursa Mini (via mini-jack to BNC cable), and a Ambient ACN-TL Tiny Lockit connected to the 552. We recorded in ProRes 4444 XQ mode at 4.6K.

As you can see in this photo, there's a bit of a delay between the Ursa Mini and the Lockit as screen on the screens. I didn't notice this until after, but I'm not sure why the camera has the ';' in the timecode, as both the camera and incoming timecode was 25.00fps, and we definitely had the 'Time Code Drop Frame' set to OFF. In retrospect, I wish I also took a photo so you could see the timecode icons clearly on the camera - opps!



Here's a screenshot of how things look with the camera audio and the 552 audio at the exact same timecode point:



As you can see, the audio is obviously quite out of sync! We then got Premiere to re-sync the audio based on the audio, which gave us this:



It's closer... but it's still not spot on. Here's a look zoomed in 100%:



As you can see there's some phasing issues, as it's still off by a bit. We then put an 'Invert' effect on the camera audio, re-exported it and brought it back into Premiere and manually lined it up - which you can see here:



Unfortunately, we have to return the camera today, but any questions let me know!

Chris, great quick review. Picture 4 shows audio that looks very close in its sound, but out of sync as you mentioned. That does look trivial to resync manually in post. Picture 5 demonstrates that a manual sync can be superior to what Premiere did at least.


Sent from my iPad using Tapatalk

Chris, could the audio sync be off a bit due to the audio 48KHz word clock being slightly off? I know external,audio recorders I used to use with video had both a time code input and a word clock LTC inputs for the audio, to lock the system to a master clock. This kept everything in sync. If the 49KHz clock drifts a bit, the audio can suffer a little bit and might drift out of sync. What do you think?
Cheers

That does look trivial to resync manually in post.

Depends if you have a guide track or not. If you're just relying on timecode and don't record any audio - then you wouldn't have any other audio reference anyway. I guess if you're just replacing the camera audio with external audio then the phase issue doesn't really matter - and if you get within a frame accurate, that's probably good enough. I guess at this point, it's probably just easiest to sync via audio using Premiere/FCPX/Resolve/PluralEyes, and just have the timecode as a backup.

Sadly, I didn't have enough time to test recording the timecode to the audio as per John's suggestion - next time!

Chris, could the audio sync be off a bit due to the audio 48KHz word clock being slightly off?

Probably! It seems the timecode is always going to be a few frames off - probably because the Ursa Mini is doing so much grunt work processing all the 4.6K footage, and simply can't keep up with the audio.

If I had more time, it would have been good to test the Ursa Mini's Reference input as well, syncing up to something like a Sound Devices 744T.

I only noticed this after the fact, but it's also a bit odd in the photo of the screen showing Timecode on the camera that it says "09:15:32;00" - as if it thought it was somehow doing drop-frame timecode (which is odd considering both the TC input & the camera were set to 25fps - and the 'Time Code Drop Frame' button was DEFINITELY set to OFF.

To quote the manual:

This input automatically recognizes and switches between timecode and reference input signals. Synchronize Blackmagic URSA Mini to a common reference signal, such as tri-level sync, by connecting to the ‘reference’ BNC input marked ‘REF In’. This lets you sync URSA Mini to other SDI video equipment, for example, when using multiple cameras connected to a switcher. You can also use this connector to record timecode from other professional audio and video equipment, such as audio mixers and clapper boards. This ensures audio and picture, or video from multiple cameras, can be easily synchronized during post production.

As there's only one physical input, I GUESS you'd have to sync the tri-level sync first, disconnect the cable, then hook up the timecode input and leave it connected. Every time you change batteries or do a reset, you'd probably need to re-sync the tri-level sync, which would get annoying.

Something to test next time we hire out the camera... or buy one.

Useually TicColor Sync, like "Black Burst" is an analog BNC input, not a SDI in, and T/C in is a digital connection, bit can be analog also.
Cheers

Thanks to the awesome guys at Empire Rentals, we've been playing with the Ursa Mini 4.6K and the brand new Public Beta 4.0 firmware yesterday and today, testing things out before we press the "buy" button. Our team did a bunch of camera/lighting tests overnight, but the main thing I personally wanted to really focus on is the audio performance. First up, I have to say I'm very impressed. It's LIGHTYEARS better than the Blackmagic Cinema Camera, Blackmagic Production 4K and Pocket. It's no Alexa - but that's to be expected considering the price difference!

Here's some quick observations:

a) The Ursa Mini doesn't have any of the frequency response issues that the original BMCC and BMP4K cameras do, which is awesome.

b) The Ursa Mini is really good for single operator jobs - as you can hook up microphones directly to the camera, and adjust levels on the fly.

c) There does seem to be some phasing issues, but they're easily fixed in post.

d) You can send timecode to the camera, which is awesome - but there does seem to be a delay between picture and audio, so you'll need to do an off-set in post. The annoying part though is that it's samples off - not frames, so you can't just have a frame off-set.

e) The headphone output is much better than previous Blackmagic Cameras in terms of noise level, however it's still not perfect. There's also a massive audio delay on the headphone output, which makes monitoring off the camera almost impossible. Personally, we'll just always hook up an external mixer to the camera, and monitor off that, and only switch over to the camera's headphone output for confidence checking, and calibration. If the camera operator had headphones in, they'd go insane.

Here's a bunch of unscientific tests we did in the office (i.e. not in a sound proofed studio or anything fancy):

1) Our first test was to play our previously made "test" track on our Sound Devices 552, connected via XLR and record it in the camera. This is basically just a quick and dirty test to see how the frequency response of the camera performs. The camera was in RAW mode, the audio input was set to XLR, both inputs were set to Line with a 86% gain. Essentially we just save our "test" track on the 552, press play on the 552, then record on the camera.

You can download the files here:

- Test Track
- WAV file straight from the camera

Here's a screenshot of the test track:



Here's a screenshot of what came off the camera:



As you can see - they're pretty close, which is much better than our previous BMCC tests!

2) The second test we did we again had the 552 connected to the Ursa Mini via XLR. On the left channel we had a Sanken CS3e and on the right channel we had a Sennheiser ME66. Again, the camera was in RAW mode, the audio input was set to XLR, both inputs were set to Line with a 86% gain. We also recorded to the 552 for comparison. Apologies in advance - I'm certainly not a voice over artist!

You can download the files here:

- WAV file straight from 552
- WAV file straight from the camera

3) The third test we did was just plugging a Sanken CS3e and ME66 directly to the camera via XLR. We had the audio input set to XLR, with both inputs set to Mic, both channels set to High with Phantom Power on, the Channel 1 gain set to 86%, and Channel 2 gain set to 96%.

You can download the file here:

- WAV file straight from camera

4) We then did a similar test with a Sanken CS3e and a COS11 wireless lapel via a Sennheiser G3 kit hooked up directly to the camera. This time I fired myself as the voice over artist, and got Nick to take over because he sounds so much nicer than me.

You can download the file here:

- WAV file straight from camera

5) Finally, we did a Timecode test. We had a Tentacle on the Ursa Mini (via mini-jack to BNC cable), and a Ambient ACN-TL Tiny Lockit connected to the 552. We recorded in ProRes 4444 XQ mode at 4.6K.

As you can see in this photo, there's a bit of a delay between the Ursa Mini and the Lockit as screen on the screens. I didn't notice this until after, but I'm not sure why the camera has the ';' in the timecode, as both the camera and incoming timecode was 25.00fps, and we definitely had the 'Time Code Drop Frame' set to OFF. In retrospect, I wish I also took a photo so you could see the timecode icons clearly on the camera - opps!



Here's a screenshot of how things look with the camera audio and the 552 audio at the exact same timecode point:



As you can see, the audio is obviously quite out of sync! We then got Premiere to re-sync the audio based on the audio, which gave us this:



It's closer... but it's still not spot on. Here's a look zoomed in 100%:



As you can see there's some phasing issues, as it's still off by a bit. We then put an 'Invert' effect on the camera audio, re-exported it and brought it back into Premiere and manually lined it up - which you can see here:



Unfortunately, we have to return the camera today, but any questions let me know!

Using update 4.0, I managed to jam two Ursa Mini 4.6 K cameras to the external free run time code of a Zoom F8 audio field recorder. It was easy. We set the first one by connecting the Zoom to the UM46 ref in tc in and repeated it for the second camera. Both cameras immediately reflected the new timecode. I did notice that the code was 3 frames off from the Zoom on both cameras. This variance has been reported as something BM will address in a subsequent firmware update. Regardless we are moving in the right direction.
Shout out to Chris Hocking and all the great posters in this forum. Your help has been invaluable as we navigate uncharted waters in the world of these great new digital cinema cameras. Thank you!



Sent from my iPhone using Tapatalk

I have the same issue with audio sync. It never used to happen on v3 of the camera software. Real pain when I bulk upload videos straight to web...