GH1 Stereoscopic Rig Collaboration

[g.l]

Resistance is measured between pad 3 and ground by the supply of a small current that flows between those two pins.

Sure, but at which pole is the resistance actually measured? To my mind it has to be measured after the resistors in the remote, ie. at pole 4 / ground. Or am I wrong about this? Could the same pole that sends out the current (3) also measure the resistance?

 

[David Cole]

Sure, but at which pole is the resistance actually measured? To my mind it has to be measured after the resistors in the remote, ie. at pole 4 / ground. Or am I wrong about this? Could the same pole that sends out the current (3) also measure the resistance?

Same pole supplies power and reads the resultant drop as a resistance value. Just like a multimeter measures resistance with only two probes.

 

[crunchy]

I assume (yeah... I know what assuming does) that it works like this:

tip - R stereo +
pad 2 - L stereo +
pad 3 - remote resistance
pad 4 - ground

What is pad 3 and what is pad 4?

I assume that pad 4 is the case. If this is true, then:

pad 3 = GND
pad 4 = remote resistance

At least pad 4 is at higher potential than pad 3 and potential of pads 1 and 2 is at the same level to pad 3.

http://www.instructables.com/id/Panasonic-G1-GH1-Remote-Shutter-Release/

So, GND is the signal connected to 36K resistor (at least my measurements suggest so). Please, let me know if I am wrong.

 

[David Cole]

Seems that you are right. Found this at DRPReview....

"the jack in the camera actually uses four conductors (like a 2.5mm cell phone jack) but the fourth connector is used only for the remote-control cable; trickery via resistors is used to operate both the AF-enable and shutter-trip functions from the third and fourth barrel contacts only."

So:

tip - R stereo +
pad 2 - L stereo +
pad 3 - ground
pad 4 - remote resistance

In any case, no worries connecting the "pad 3" from each camera together.

 

[crunchy]

Good.
Now I have a little more time to continue with experiments.
My preliminary tests showed that the cameras are mostly (let's say more than 90%) synched within 1/500s (I haven't measured the synch with higher precision, yet). This is usually enough for 1/50s shutter speed.
Will post results when I get more data.

 

[David Cole]

Crunchy:

Have a barrage of questions for you (as you are the second datapoint that we have here):

1. How are you characterizing your sync offset? Wimmer's CRT sync tester?
2. Do you know the initial offset in ms? Do you see significant drift in the same test run, after 5 min?
3. Is the offset the same, or close between power-cycled test runs?
4. How close are your serial numbers?
5. Are you using manual lenses (guessing yes on this)?
6. How are you powering up both cameras simultaneously? Simple SPST switch (like me) or something fancier and less likely to bounce?

Just as a point of reference, my rig has consistently less than .02 ms of drift at the beginning of each test run and seldom more than .1 ms after 5 min. My GH1s have sequential serial numbers. I'm using the circuit for power-up and record trigger that is described in my earlier posts - nothing fancy. This result is only achievable for me with manual lenses attached. I'm measuring with both the Wimmer CRT test and a NTSC blanking sync tester looking at the composite video outs. The results mostly co-vary from those two tests. Sync is so reliable that I do not check in the field at all... I just power-cycle and shoot. Shooting basketball at the moment (720 60 w/ 120 shutter) and seeing 0 phase offset.

Thank you!

 

[g.l]

Thanks guys, still waiting on my bodies to try it.

In the meantime, some good news on the GH2 monitoring front: turns out the HDMI output is tapped off the LCD feed. If the LCD is rotated 180deg, it turns the image upside-down AND mirrors it horizontally - and it outputs exactly the same on HDMI! So for a two-display/half-mirror solution, we no longer need a screen (or controller) that can mirror the image.

I also found a really small HDMI LCD controller, but they haven't confirmed yet if the 5.6" 720p panel is compatible. Is anybody interested in using this panel?

 

[crunchy]

1. How are you characterizing your sync offset? Wimmer's CRT sync tester?
2. Do you know the initial offset in ms? Do you see significant drift in the same test run, after 5 min?
3. Is the offset the same, or close between power-cycled test runs?
4. How close are your serial numbers?
5. Are you using manual lenses (guessing yes on this)?
6. How are you powering up both cameras simultaneously? Simple SPST switch (like me) or something fancier and less likely to bounce?

1. No. I just built up some "hand-held" device with LEDs. However, I'll have to check it with Wimmer's tester as well. The monitor was in the attic and it is so cold that I'll have to wait till tomorrow to be sure that it won't die due to condensation...
2. The initial offset does not seem large. It is below 1/500s. Note that I tried with two Panasonic lenses (14-140 and 14-42) and with mixed lenses (14-42 Panasonic and Nikon manual). Drift is very large. About 1ms/min. Will have to change the oscillator(s) one day.
3. Usually it is the same. However, I'll have to do more rigorous tests, first (with CRT monitor).
4. Serial numbers are VERY different. They are about 3/4 of the year apart (even more, one is black and the another one is red ).
5. No, mixed or both Panasonic (not the same type), as already said.
6. The switch is the same as in your case. I had no time yet to make some improvements. Still I am planning to do that.

Plus: One camera is GH13, the second one is GH1 (still did not have time to hack it).

... This result is only achievable for me with manual lenses attached. I'm measuring with both the Wimmer CRT test and a NTSC blanking sync tester looking at the composite video outs. The results mostly co-vary from those two tests. Sync is so reliable that I do not check in the field at all... I just power-cycle and shoot. Shooting basketball at the moment (720 60 w/ 120 shutter) and seeing 0 phase offset.

I'll try with two Panasonic 14-42mm when I get the second one and will report about my findings (waiting for delivery).
Please, can you tell me in more details what did you mean by "The results mostly co-vary from those two tests"? Can you give me an example when those tests do not co-vary (does it mean that NTSC tester gave excellent result, but the actual synch with Wimmer method gave bad result - let's say several ms missynch)? What kind of "NTSC blanking sync tester" are you using? The same as with Rob Crockett? What kind of experiment are you performing? I am testing with oscilloscope (connected to composite video outs) and at the same time am checking video synch (with my device). However, due to very low time resolution, I can't be quite sure about my findings.

 

[crunchy]

2. Do you know the initial offset in ms? Do you see significant drift in the same test run, after 5 min?
3. Is the offset the same, or close between power-cycled test runs?

After first tests with Wimmer's synch and CRT monitor, the results are the following:
Initial missynch is about 0,8 milliseconds. Red GH1 is always faster.
Drift is about 0,8 milliseconds/minute. (CORRECTED previous results!)
Lenses: 14-140 + 14-42, In auto focus mode, but in Manual video mode (since brightness is changing all the time due to the test procedure), 1/50s shutter, 75Hz monitor (yes, you can still see some brighter and darker bars!).

I can see some correlation between composite synch and the actual synch. If composite synch is not good, the actual synch is also bad.

It's over the midnight and bed is waiting for me.

 

[crunchy]

David, can you describe in more details the experiments you were performing with Panasonic lenses? You said that you had problems with synch when having "auto" lenses. Can you describe your results in more details? How many bad synch shots have you got (relatively)?

My set-up is the following: GH13 and GH1, Panasonic lenses 14-42mm on both cameras, AF mode, Movie S mode (1/50s).
I switch on the cameras simultaneously (delay between power-ups can be adjusted by the uP). In fact I power-up the right camera 1.2ms later in order to get initial missynch about -0.75ms. Due to drift, the missynch becomes 0ms after 1 minute and +0.75ms after 2 minutes.
After power-up I wait until "memory card writing" signal vanishes. Then I start shooting. Yesterday, after tinkering with micro-controller-based synchronizer (let us call it "3DSLR Master") I couldn't get bad results (after more than 20 attempts). It's true that I spent hours and hours in experimenting in order to find exact timing within 3DSLR Master. However, it seems that I'll probably spend additional hundreds of hours for fine tuning. Namely, it seems that it is crucial to choose right timing for introducing "trigger" signals to the cameras.
The device (3DSLR Master) displays disparity continuously and according to my findings the shown numbers are in exact correlation to the actual disparity (when checked with synchtester.exe on CRT screen), even during actual shooting.

g.l., have you already got the "bodies"?

 

[g.l]

g.l., have you already got the "bodies"?

No. I think the shop I ordered them from is messing me around, need to chase it up this week.

 

[David Cole]

My GH2s are on order from Amazon with a March 10 - March 25 estimated delivery.

 

[David Cole]

...what did you mean by "The results mostly co-vary from those two tests"? Can you give me an example when those tests do not co-vary (does it mean that NTSC tester gave excellent result, but the actual synch with Wimmer method gave bad result - let's say several ms missynch)? What kind of "NTSC blanking sync tester" are you using? The same as with Rob Crockett?

The correlation of my NTSC sync (yes, I'm using Rob's Sync Shepherd circuit, modified to increase sensitivity at the low end and drive a numeric display) to the Wimmer CRT sync isn't 1:1. IF the NTSC sync is badly off, the Wimmer sync would tend to be off - but - not always. But, since I stopped using M4/3 lenses and went with lenses with no electrical connection to the camera, all of the mis-sycn problems ended.

 

[David Cole]

... Serial numbers are VERY different. They are about 3/4 of the year apart (even more, one is black and the another one is red ).
5. No, mixed or both Panasonic (not the same type), as already said.... Plus: One camera is GH13, the second one is GH1 (still did not have time to hack it).

Wow. That's a lot of potential causes of variability to try to isolate!

 

[David Cole]

David, can you describe in more details the experiments you were performing with Panasonic lenses?

Yes. I need to be clear about this. I could not - in over 100 tests - achieve sync that could be relied on every time with M4/3 lenses that have an electrical connection to the camera. I believe that some non-deterministic delay is imposed during camera startup, when it initializes/calibrates the drive motors in the lenses. You can hear it happen when the cameras startup. Also, you can see that focus changes after a camera power-cycle - even if you have autofocus off (which I ALWAYS did).

After I gave up on M4/3 lenses, I found initial sync and drift over time to be VERY acceptable.

 

[crunchy]

Wow. That's a lot of potential causes of variability to try to isolate!

But still the results are not bad. In meantime the second camera became GH13 and of course, I am now using two Panasonic 14-42mm lenses (the worst, but the cheapest :happy: ). I just added voltage stabilisers (custom designed) which will allow me to use external batteries without any problems. Now everything is controlled by the micro-controller (power-on-off, sending focus and trigger signals to two cameras) and the missynch are written on an LCD display with resolution of 10 microseconds.

I can't understand that your measurements of video missynch and the actual missynch do not correlate. After many testings I've found that they are correlated without exceptions. Maybe I haven't done enough tests or was not precise enough?!? Can you give me one example what is (in your opinion) considered as non-correlated (the difference between Rob's device and Wimmer's screen)?

I still haven't found the appropriate case, but when I get it I'll send a picture.

What kind of beam-splitter are you using?

 

[David Cole]

How much resolution do you have on your video sync measurement? I was in the .005 ms range with my video sync tester. Perhaps the jitter in the sync tester itself was a factor for me. Gross mis-sych errors did mostly correspond between the two testing methods. I have a correlation table somewhere..... I'll try to dig it up for you. {by the way, I also tested twin Sony HDX-CX550v cameras with a Lanc-Shepherd. The Lanc-Shephard and my video sycn tester were only roughly correlated as well}

RE: Beamsplitter ... On the testing bench I have a 3DFilmFactory rig. I have several custom rigs that I actually shoot with. I just ordered a Hurricane Rig from Genus... looks like a really nice piece of kit! If you are in the market, I hear REALLY good things about this kit: http://cgi.ebay.com/3D-RIG-3D-CAMER...ltDomain_0&hash=item3364570ce6#ht_7588wt_1141. It's basically the same as my 3DFilmFactory rig for 1/2 the price.

 

[crunchy]

How much resolution do you have on your video sync measurement? I was in the .005 ms range with my video sync tester. Perhaps the jitter in the sync tester itself was a factor for me. Gross mis-sych errors did mostly correspond between the two testing methods. I have a correlation table somewhere..... I'll try to dig it up for you. {by the way, I also tested twin Sony HDX-CX550v cameras with a Lanc-Shepherd. The Lanc-Shephard and my video sycn tester were only roughly correlated as well}

My resolution is 10 microseconds instead of 5 in your case. There is some slight jitter as well. If you find that correlation table, it will be great. What is roughly correlated for you? For 1/50s shutter speed (video), synchs below 1-2ms are usually negligible. Do you agree?

RE: Beamsplitter ... On the testing bench I have a 3DFilmFactory rig. I have several custom rigs that I actually shoot with. I just ordered a Hurricane Rig from Genus... looks like a really nice piece of kit! If you are in the market, I hear REALLY good things about this kit: http://cgi.ebay.com/3D-RIG-3D-CAMER...ltDomain_0&hash=item3364570ce6#ht_7588wt_1141. It's basically the same as my 3DFilmFactory rig for 1/2 the price.

Very nice piece of work. I like it, although I'd make some changes in construction.

 

[g.l]

How much resolution do you have on your video sync measurement? I was in the .005 ms range with my video sync tester. Perhaps the jitter in the sync tester itself was a factor for me.

Actually that seems likely to me, as I can't imagine that the two video blocks could run asynchronously. At least not in a cheap-as-possible mass-market design.

Would be great of course as I could use my Panny primes .

Still no bodies. I got a really good deal on them - unfortunately the shop probably made a mistake, but rather than cancel the order, they're pretending they can't get stock. It looks like the bodies are genuinely hard to come by in the UK at the moment, but I'm pretty sure that a few bodies are floating around and they're sending them to people who're paying more. If I can't get them within 2 weeks I'll have to rethink...

 

[crunchy]

Actually that seems likely to me, as I can't imagine that the two video blocks could run asynchronously. At least not in a cheap-as-possible mass-market design.

What did you mean??

Would be great of course as I could use my Panny primes .

Exactly. I am also thinking about getting Panny primers (probably 14mm F2.5) if this approach will prove to be successful.

However, maybe my and David's measurements also differ since I have PAL settings. And, of course, one important difference is that I am using separate switches for FOCUS and for TRIGGERING shots. Before each shot I first focus (always!!), then I trigger the shot. David, can you try it with your cameras as well?