Ben B

Timecode difference using Alexa Mini and Tentacle

Recommended Posts

Ben B   

Hi!

I had a call from the editor of a feature we finished 2 month ago saying that every files were too late from between 1 to 3 frames compared to the scratch track. It was not a constant off set but a difference ranging from 1 to 3 frames.

We used 2 different set up for the different shooting phases. The first one was tentacle into a Sonosax SX-R4, and tenatcle into the Mini. Both units were constantly fed with timecode from the two Tentacles. The second set up was exactly the same but with a 633 instead of the SX-R4. This timecode difference was on every file regardless of the recorder used.

The editor told me and my colleague that even the test shoot were off by about 2 frames. We got this news 2 months after being done with the 9 weeks shooting. 

Has anyone any idea where the problem could come from? I'm not really looking for a solution to re-sync the clips but more as to where this difference could come from (and ultimately not reproduce it in the future...).

Thanks!

Share this post


Link to post
Share on other sites
rich   

ive been speaking to someone who has been mixing a show shooting on two alexa mini's.

they have been using lockit boxes jammed via ambients ACN and using the master controller connected to the 788.
edit has reported the same thing.
the assistant editor helpfully emailed the producers to say that there was the same offset.
using a digislate proved that the timecode from the lockit boxes was not the issue.
the DIT on the show said that it is a well known thing with the Alexa mini and there is a workaround - dont know what it is though.

but since the digislate was brought to set, the assistant editor agreed that the issue was nothing to do with sound, but somewhere in the camera, and now its not an issue any more. not that they told production, but thats another story.

 

it never ceases to be irritating that whenever there is a sync issue, the finger is automatically pointed at sound. almost every top end recorder has timecode accuracy to rival the best external syncboxes. how many cameras could claim the same?

Share this post


Link to post
Share on other sites

Complaining about a 1 or 2 frame offset, while interesting is pretty much OCD.

Analog Nagras were 4 frames out and the world did not end.

 

Share this post


Link to post
Share on other sites
37 minutes ago, al mcguire said:

Complaining about a 1 or 2 frame offset, while interesting is pretty much OCD.

well, with a constant offset no big deal, but if it's 1 to 3 at random I can understand that post gets unhappy. 

my shoots on a Alexa Mini (two times 4 days) were both without sync box (synched at the beginning and at lunch) and we didn't have any offset at all.

 

Share this post


Link to post
Share on other sites

I have been working with a local DIT here to track down the source of a 2 frame offset when working with an erx on the Alexa mini.  Fortunately it's a constant 2 frames, and doesn't drift, so it's easy to adjust for in the erx menu. Testing at home by syncing 2 Nomads with an ERX and clapping a slate, I do not get any offset, so I imagine it's something with the ERX and Alexa Mini combo.  I have not checked the sensor settings on the cameras I've worked with recently.

Share this post


Link to post
Share on other sites
axel   

same here, shooting on Alexa Mini, jam synced twice a day, after the first day the assistant editor complained about 1-2 frames drift (Alexa behind).  I have the Cantar X3 as a master, next tried to feed the X3 from the Lockit (ACL 204), nothing changed. Now I compensate the Lockit with +1,5 frames (just used to carry TC to the cam) and AE says sync is OK now. 

 

Share this post


Link to post
Share on other sites
mikewest   

Yes the perrenial problem!

Did a 2 camera series last year, a Mini and an Alexa

DP in his "wisdom" said no t/c boxes on the cameras

Oh dear after the first we post came in with their Ambient kits

Next 4 weeks oh dear cameras do not agree please check

Last day of shoot I used my Denecke tc boxes - no problems!

Use a tc slate and focus on SOUND

mike

Share this post


Link to post
Share on other sites

@al mcguire that would be simple if the mini accepted it without a modification, but not all mini's are modified this way.

@mikewest the DIT I'm working said he thought he remembered the denecke box & Alexa mini combo were frame accurate.  Trusty old Denecke.

It does bring up the question of what is different about the TC stream that would cause the mini to be on or off??

Share this post


Link to post
Share on other sites

I still try to make the case that on multicamera shoots the TC/Gen boxes are as much about keeping the cameras in sync with EACH OTHER as much as keeping them both in sync w/ sound....

Share this post


Link to post
Share on other sites

you don't need to mod the Alexa Mini's SDI 2 connector to be a separate Tri Level Sync input to achieve this - like the other Arris, you can choose the TC Input as the Sync source in the Sensor menu - in which case it Clocks/Syncs/Genlocks (whatever you want to call it) off the incoming TC feed from your Lock Box as well as getting the TC time from it. It might not be quite stable enough for a 3D rig (which is what they offer the SDI connector mod for) but should be fine for the purpose being discussed here.

Share this post


Link to post
Share on other sites

with most DPs you'll have a hard time trying to convince them to clock their sensor of an external device - and frankly I can't blame them (personally I would be worried about flicker etc and wouldn't risk it before running a lot of tests).

what is odd that the problem seems to be frequent enough that a lot of people have seen it, while there are also a lot of shoots where it's no problem at all.

 

Share this post


Link to post
Share on other sites
astro   
you don't need to mod the Alexa Mini's SDI 2 connector to be a separate Tri Level Sync input to achieve this - like the other Arris, you can choose the TC Input as the Sync source in the Sensor menu...


Well, that's the theory.

We tried that with a tentacle and a tiny lockit with no success. Only after we got the modification done and permanently attached a tri level lockit the inconsitant tc changed to a consistent one frame offset.

Share this post


Link to post
Share on other sites
davidm   
3 hours ago, Philip Perkins said:

Just curious--has there been testing to be sure that having the genlock source be TC doesn't affect the picture at all?

Genlocking via timecode is Arri's recommended practice. It's noted in the current Alexa Mini and Amira manuals and firmware notes.

TImecode works in this instance because the camera has a global shutter and the TC is triggering a single full sensor read of each frame rather than a traditional and more complex TV scanning sequence. The Alexa and a Mini with the genlock via BNC mod can use black burst or Tr-level sync as an alternative. 

The only caveat with using TC as genlock is that a stable, high quality generator must be used.

 

 

Share this post


Link to post
Share on other sites

In the case of the Amira, you have both options available, a dedicated tri-level sync (genlock) input or you can set it to sync from the TC input. Ambient recommend the latter, actually. I have certainly done this on Alexa Minis and Amiras on a 2 week shoot with up to 11 cameras using a mixture of Ambient ACL204s and Tiny Lockits, with no reported problems. Several other one-day one or two cam concert shoots with ACL204s only.  I recently asked in another post here if anyone had done it with Tentacles, with no positive replies, but Tentacle themselves say they have. I've noticed that it does confuse ACs that they have to set this up in two different menus. 

Share this post


Link to post
Share on other sites

Is there any recommendation from Tentacle's side as to which settings to use on an Alexa Mini, as these issues seem to happen often with Tentacles (without being their fault obviously)? Nick, you wrote they have tried syncing from the TC input, so is that their recommended way of doing it?

Share this post


Link to post
Share on other sites
6 hours ago, davidm said:

TImecode works in this instance because the camera has a global shutter and the TC is triggering a single full sensor read of each frame rather than a traditional and more complex TV scanning sequence.

The Alexa family doesn't have a global shutter (but they have a very fast readout, so the typical rolling shutter problems much smaller then on other cameras)

Share this post


Link to post
Share on other sites

Here's their reply to my thread on the Tentacle forum. They gave the same reply to the OP from this thread we are in now, in his similar thread on the Tentacle forum. Doing a bit of a web trawl about this question, the only posts I could find where people specifically had trouble with a Tentacle being used to BOTH TC lock and Genlock an Alexa Mini, it would appear the users were unaware that Tentacles do not automatically change their TC Framerate when they are jammed (from a Sound Recorder, for instance) and they were subsequently sending TC at 25fps into a camera who's framerate (project timebase) was set to 23.976. They had jammed from a recorder set to 23.976 so they thought they were good to go. They needed to manually set Tentacle Framerate using one of their control apps. This mismatch is likely to cause the facility of 'tuning' the camera's internal clock to the sync timing implied by the incoming timecode (which is how Arri and several Audio Interface manufacturers do this one cable TC and Sync trick) to fail miserably - no matter if the source was an Ambient, Denecke, Tentacle, or Mozegear product. Interestingly Tentacle still think there will be a one-frame offset with the camera genlocking off timecode, it'll just be more stable as regards Drift over a longer time. This suggests it's actually caused by the processing time inside the camera of the TC data (not uncommon). They just recommend setting an offset in the camera as a matter of course.  My regular clients don't question static offsets, even of several frames. Drift, on the other hand, given they're usually editing in a 'Multicam' workflow and cutting to music, is a big issue.

Screen Shot 2017-03-30 at 7.27.49 pm.png

Share this post


Link to post
Share on other sites
davidm   
19 hours ago, chrismedr said:

The Alexa family doesn't have a global shutter (but they have a very fast readout, so the typical rolling shutter problems much smaller then on other cameras)

I stand corrected regarding the roller shutter on the Arri's, but I'm told it's damn near a global shutter. 

Share this post


Link to post
Share on other sites

I remember that Lockits/Clockits need to be re-tuned after a certain time, using the Ambient Clockit Controller, because any crystal drifts a bit over longer time periods (we're talking months to years here). Does anybody know if this is still true for "blue" lockits and ACN, and how it works with Tentacles? Do Tentacles tune themselves when connected to the app and GPS?

Share this post


Link to post
Share on other sites

As Ambient explained to me last year...with Blue ACN series lockits, if you are in ACN mode, the slave lockits will kind of tune themselves in real time to the master unit in a set (but only for the time they are being used). Therefore tuning a set of lockits before a project is less important (even if some of them are hired in, and may have been tuned to a different reference), but Ambient still suggest tuning each one to a known reference (i.e. GPS or one of the Lockits) every year or so, which makes each pack need to 'deviate' less using ACN.

Share this post


Link to post
Share on other sites
Klaus   

@ Nick: This. Thanks for jumping in (I ashamedly admit being labelled a Tentacle thread I haven't come across this  before ).

Anyway, tuning goes back a long way. The Controller came out in 1994 (the first Lockti boxes 1992-late 1993 only offered manual tuning). This was a recommended process every couple of years to maintain relative accuracy - and this has been completely replaced on "blue generation" with ACN. In C-.Jam slaves will sync up within 10uS deviation from their Master. Still, it doesn't hurt to start with a close as possible reference, but I wouldn't recommend getting a Controller if you are ACN equipped (especially with what we have in pipeline) - the only advantage a ACC501 still would give you ATM is tuning to absolute reference via GPS (what we do in manufacturing). But that isn't something that couldn't be done over ACN in general ;).

[/OT]

Also, Nick has posted all the relevant settings on the ALEXA Mini as per Tentacle's customer support. We additionally invite everybody (also Tentacle users ;) ) to benefit from our video tutorial section www.ambeint.de/en/university. Once these settings are observed it doesn't matter what brand of TC generator you use, the Mini wil do just this: regenerate the TC to the external source and follow it. Neither of the products in mention will introduce varying offset or drift as long as it's stable and sync with the other equipment (which I am confident to certify for our products) and configured correctly.

That 2 frame offset has been repeatedly experienced and reported in the field with the Mini but could not be reproduced in house with ARRI. ARRI support told me they would appreciate direct feedback to isolate the issue.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now