GH1 Stereoscopic Rig Collaboration
- Thread starter David Cole
- Start date
David Cole
Well-known member
Agreed. There is a risk of mechanical bounce but as the power switch is single throw, it is unlikely to affect sync.
What would you recommend ? An SSR might induce noise.
Mechanical bounce can, by itself, introduce some missynch due to slight differences in inner electronics, although it should not be large. On the other side, have you measured (on oscilloscope) the difference between the power signals on both cameras (when you power them up?)? I'd also be interested in the measurements.
What level of noise would be induced by the SSR? Have you tried them yet?
David Cole
Well-known member
We are using the free Camera Sync Tester application from 3DTV.at (http://www.3dtv.at/Knowhow/Synctest_en.aspx).
We simply run the application on a PC with a CRT display and point the camera rig at it. To analyze, we bring each file up on a separate display in Quicktime (which offers a frame accurate counter) and look at the same frame, in the left and right camera files. The following conditions are observed:
1 - perfect sync - the scan line of the CRT illuminates the same position on the recorded chart with both cameras.
2 - frame slip - the cameras are in sync, as per above, but the record began later in one camera or the other. This is not a problem, you can easily fix in post.
3 - bad sync - the scan line is in very different positions on the recorded charts.
By the way, to lay down a sync check in the field, you can use a diffused photo strobe set for fast exposure. Just fire the strobe at the cameras (make sure that the flash is diffused or if could damage the image sensor). You can count the scanlines in post to see if the flash was recorded at exactly the same position in both cameras.
David Cole
Well-known member
Yes. We've measured the voltage increase that occurs on the remote line (pins 3&4) when the cameras power-up. (Presumably, the GH1s are reading the remote resistance values with this)
This DSO image shows the power-up event from both cameras (at 2ms / DIV)
The events seem to occur simultaneously every time (at least as far as this 150MHz DSO can resolve).
We've moved on to measuring the pull-up of the data line on the GH1's USB output. I'll post those results next.
g.l
Well-known member
Did you try without memory card installed?
David Cole
Well-known member
g.l
Well-known member
The idea is to boot up first, then insert the cards, in case the card access during boot time is causing the unpredictable sync shift.
It's conceivable that a global frame capture timer is initialised only after the cards are accessed for the 1st time, so this access might cause the random delay. Worth trying.
It's conceivable that a global frame capture timer is initialised only after the cards are accessed for the 1st time, so this access might cause the random delay. Worth trying.
David Cole
Well-known member
OK... We'll give it a whirl. Perhaps we'll open a testing WIKI at beampath3d.com to track tests and results. There are a LOT of variables to test against sync and it would be good to organize the test results.
Also, narrowing in on an electronic measure that correlates with sync (e.g. some wiggle of the USB data line) will probably take some heavy testing as well.
Also, narrowing in on an electronic measure that correlates with sync (e.g. some wiggle of the USB data line) will probably take some heavy testing as well.
g.l
Well-known member
It would also be interesting if you could compare the sync in live-view before the take, with the sync of the take itself. That would tell us if the capture timer is global, or reset for every take.
Does shooting the CRT give you a steady-ish bar that you can compare visually on both cams prior to the take?
Does shooting the CRT give you a steady-ish bar that you can compare visually on both cams prior to the take?
David Cole
Well-known member
David Cole
Well-known member
Crunchy has suggested that we relocate this thread to, "GH1 Hardware, Lenses, etc."
Anyone know how?
Anyone know how?
g.l
Well-known member
I think we should keep it here to remind Vitaliy (or others once the hack is more opened up) that a firmware solution is still of interest. I actually mentioned syncing to Vitaliy a while back and he was interested in it, but it understandably it's taking a backseat to bitrate etc. hacking. But hopefully it's still in the queue.
If they are consistent then sync is likely global and unaffected by takes (apart from an overall gradual drift that's probably unavoidable).
I've never used a GH1 live - so the LCD is not running at full capture rate? If it did, then just pointing it at a CRT may show a steady bar to compare. It wouldn't need precise measurement, rather you could see if the bars are always the same distance apart between takes.
If they are consistent then sync is likely global and unaffected by takes (apart from an overall gradual drift that's probably unavoidable).
David Cole
Well-known member
I'll try to look closely at the LCD during the next test (and maybe even snap some picts). Our precursory analysis was the display wasn't helpful. This could be because the scanline position jumps around from frame-to-frame (unlike a genlocked camera / display).
Another challenge is that, mounted on a mirrorbox, the LCDs are some distance apart and not orientated the same way. This makes visual comparison difficult.
Another challenge is that, mounted on a mirrorbox, the LCDs are some distance apart and not orientated the same way. This makes visual comparison difficult.
g.l
Well-known member
Right, I guess the challenge is to match the CRT refresh rate to the cams.
A mirror might be useful. If you can snap everything visible it would be easier to analyze later in eg. Photoshop.
BTW thanks a bunch for running these tests.
David Cole
Well-known member
It's a pleasure.
By the way, if we get sync working, Aquavideo has committed to building a very flexible GH1 rig. They are developing a mini mirrorbox that can fit into an underwater housing, on a shouldermount or on a steadycam mount. I'll ask Mike Hastings at AquaVideo to chime in and drop some details.
David Cole
Well-known member
What is the proper procedure for suggesting a patch? I left a note in the " Emotions, crazy requests, donation messages go HERE" thread, but that somehow seems off-the-radar.
I guess we need to know what we're asking for. If there is any way for the camera to, once fully initialized, send a smoke signal out one of the outputs (e.g. wiggle the data line on the USB port) we could use a PIC to measure the offset. It would then be easy to get the same level of sync functionality that you get from a LANC Shepherd. The rig is in sync on power-up or else you power cycle and try again.
Can you still tell us what is bad synch? In other words, what is max. disparity which is considered as "perfect synch"? I'd like to know what are the most requent results obtained with a pair of GH1s. "Bad synch" shown on your second picture could still be used in some cases for video (I suppose that missynch was below 2ms).
g.l
Well-known member
As I mentioned, I think you can already measure the offsets from the vsync pulses on the composite outputs - they're available until you're recording right?
Re. patch request, it's hard to say anything concrete unless we know how sync could be achieved (your tests should help us narrow it down). Something like:
a) a way to tightly sync frame capture via two cameras. eg. a way to increase shutter release timing (if that would help), or a way to synchronize the boot up sequence (if that's the issue), or a way to synchronise sensor initialisation at record start (if that's the issue), or a way to provide 'software genlocking' in real-time via two-way comms (if possible).
b) a way to synchronize exposure, white balance, zoom and focus (or as many as are possible), ideally continuously in real-time, or at least via some trigger (this requires two-way comms).
Re. patch request, it's hard to say anything concrete unless we know how sync could be achieved (your tests should help us narrow it down). Something like:
a) a way to tightly sync frame capture via two cameras. eg. a way to increase shutter release timing (if that would help), or a way to synchronize the boot up sequence (if that's the issue), or a way to synchronise sensor initialisation at record start (if that's the issue), or a way to provide 'software genlocking' in real-time via two-way comms (if possible).
b) a way to synchronize exposure, white balance, zoom and focus (or as many as are possible), ideally continuously in real-time, or at least via some trigger (this requires two-way comms).
Yes. We've measured the voltage increase that occurs on the remote line (pins 3&4) when the cameras power-up. (Presumably, the GH1s are reading the remote resistance values with this)
This DSO image shows the power-up event from both cameras (at 2ms / DIV)
The events seem to occur simultaneously every time (at least as far as this 150MHz DSO can resolve).
We've moved on to measuring the pull-up of the data line on the GH1's USB output. I'll post those results next.
Simultaneously every time even though there are some "missynched" results? Hmmm... It seems very good. So power-up voltage is probably OK. Have you also measured the difference in voltages when "focus" and "shoot" buttons are pressed (although I can't imagine that this delay can cause "missynch", except frame slips).