Sony PMW-EX1 CMOS FAILS Strobe Test.

ex1lead.jpgFreshDV shot some interesting footage with the Sony PMW-EX1 and video of a police car with numerous strobes firing clearly demonstrated that Sony’s new flagship SxS camcorder, despite being knighted with the CineAlta badge, clearly has problems with capturing partial-second instances of time.

There is going to be trouble in Teaneck

The footage shot by Matthew Jeppsen on FreshDV clearly shows all kinds of half, and partial illuminated frames when shooting video of a police car with strobes that do fire at very precise intervals. There’s even indication that the camera completely missed some of the strobe flashes because the frames that are illuminated, are not in any even pattern.

lightbar.gif

If you take white strobes firing at precise, regular intervals, and shoot it with a camera recording video frames at precise intervals, then you should have a regular pattern of flashes, not a seemingly random series of flashes. This is over and above the numerous partially illuminated frames that clearly show the camera is just not doing things right.

Here’s my collection of several top strobe flashes in succession.
Click on the image for a larger version:

sonyex1failsstrobetest.jpg

Is it the CMOS imager? Processing? Something else?

It’s really hard to say. Even if the CMOS imager is “scanning” or using a “rolling shutter” one would expect the scanning pattern to be exactly the same for every frame. And if every frame of video happens the same way, at the same interval, and you point the camera at a strobe firing at a regular interval, you would expect a very even pattern of “mistakes” to occur.fx1.png
Not a completely random pattern.

Something is clearly amiss here.

I’ll be holding on to my three CCD HD camcorder till they can fix this horrible problem.

.

Advertisement

14 thoughts on “Sony PMW-EX1 CMOS FAILS Strobe Test.

Add yours

  1. My Canon HV20 has a similar issue. If the flash of a still camera is set off, it only is seen in the bottom half of the frame. The “whole” frame is never completely flashed.

  2. unless the strobe and the camera are set to sampling frequencies that are factors of each other, the periodicity you observe can be quite long. for example, turn on two strobe lights: one set at 100hz; the other set at 101hz. the pattern you observe will not repeat for 100 cycles.

    as you noted though, the partial exposure is a result of the rolling shutter scan, and this could certainly be a problem for some people, not so much for recording flashes, but to keep the image unwarped for handheld-style shots and vfx work. in those cases this camera is not the best choice. :)

  3. I see what you mean about the long periodicity.

    You also bring up a good point for VFX work, because even motion within the frame may be improperly rendered- and then when you try to augment that footage with computer rendered images that do not have rolling shutter distortion, then you have a bad mismatch of imagery. Makes you wonder about the CineAlta badge, and other higher end camcorders that also use CMOS imagers.

    Anthony

  4. After videoing my first wedding with the EX 1 last week and looking at the footage I’m very pleased to find that the EX 1 handled photographer’s flashes much better than the FX 7, which I refused to use for weddings. The XDcam codec handled the video distortion of the flash better and mostly the flash only wrecked one frame or two, making slo-motion easier to edit for weddings.

    Overall the image was top notch.

    my 0.02 cents……Vaughan

  5. Doesn’t seem like such a good test. Any strobes probably are firing faster than the frame rate the camera was shooting at and would not be fully in sync with the video camera causing half frames etc. You didn’t show us what a CCD camera was doing to the same footage. How did the footage play at full speed not stills? If I was shooting a scene for a movie with Cop car lights flashing I wouldn’t expect the flashes to be in sync with the camera all the time. My concern would be other artifacts like streaks. I would rather see if somone was shooting behind the scenes at a still shoot with strobes going off to see if there is something to be concerned about

  6. Daniel,
    It isn’t my footage, it’s at FreshDV. I grabbed frames to illustrate the point. You are right, that it would be nice to see two cameras, like an FX1 and the EX1, set next to each other shooting the same police car and compare the footage.

    But, thinking back to the still image days, a flash is some 1/50,000 of a second. A frame of video is 1/24, 1/30, 1/60 for a field of interlaced video. There’s a lot of “time” in that 1/24 of a second to fire the strobe and see the whole image illuminated.

    Like the animation of a film shutter on Wikipedia:

    The only time you can get partial frames is when the edge of the shutter passes in front of the gate. It seems that, with CMOS imagers, the amount of time where the “gate” passes down the face of the sensor (the rolling shutter) is much greater than you would think.

    The whole image may be exposed for 1/24, but it looks like they are including the time where the “shutter is rolling” across the image.

    What I don’t understand is that, how the finished frame includes both illuminated and un-illuminated parts of the image. This can only happen if the rolling shutter rate is different than the finished data frame rate. So a single frame may contain imagery from the top have of one scan (without flash) and the bottom half of a different scan (with flash) which really makes no sense to me at all.

    Anthony

  7. I’ve used the EX1 on several shoots,both outdoors on a bright and sunny day, and at night in low light conditions. I tried my best to ignore the serious problems that this camera has with moving subjects (especially fast moving subjects) but I have to finally admit that even though this camera produces stunning images, it is a horrible camera that just does not deserve the Cine Alta badge. This camera just cannot handle fast moving objects. In fact even when you move the camera a little bit, everything gets blurred. They did everything right with this camera except the CMOS. I just wish it were CCCD instead. I will be selling my EX1 as soon as possible and for a very reasonable price.
    I’m hoping that Canon comes to the rescue with something new.
    David J Rathbone
    rathboneindustries@gmail.com

  8. The trouble may be that Canon holds several patents on CMOS and is one of the developers of the technology. It’s very likely that Canon’s next big step will be to CMOS.

    We can only hope that they can but in the right processing in their Dig!c chip that it eliminates the artifacts that people find objectionable after decades of CCD use. (Remember Tube cameras were also progressive scanning- but usually viewed on progressive tube displays, so the end naturally corrected for the source since they both worked the same way.)

    Also, be sure you’re not using one of th eslow shutter settings. The EX1 can actually shoot 720p60 progressive at 60 frames a second. 1080i30 is 60 fields a second. You can also set the shutter higher than that, so there are ways to mitigate motion blur.

    If you have still images you want to send in as examples I can attach them to this thread.

    Anthony

  9. hi I test with strobo my ex1 and this problem is only with seting shutter.TLCS just set speed 0+- and ON auto shutter 1/250 working fine

  10. I was thinking of buying an Ex1 as my first serious camera. Thank god for forums like this one. I am now at a loss for what camera to buy. Maybe I should stick to something tried and tested like the Z1.

    Jeff.

  11. Jeff, I wouldn’t be so confident of the information posted on this (God hasn’t contributed). The EX1 is a fine camera and these theoretical “analyses” tell you something for consideration, but nothing more. If you are thinking of buying a camera, as always hire one and test for yourself under conditions pertinent to your use. CMOS does have rolling shutter characteristics that may cause problems in particular circumstances. CCD has other characteristics that can also be problematical. I’ve tested the EX1 for rolling shutter effects and they are minor for most serious use. In other words, repeating a test claimed to produce “disastrous results” has failed to demonstrate a practical problem; you must differentiate between the approaches of bug-finders and those actually making films. By all means buy a Z1. Cheaper, but by a long way inferior (but no rolling shutter). Or buy Rathbone’s EX1 at ‘a very reasonable price’.

  12. Incidentally, if you find moving the camera blurs the image more than you expect, turn off detail. Many users prefer to set detail to ‘off’ and do all sharpening, if necessary, in post. Putting the matter in its crudest terms, the sharpening routine hibernates when you wave the camera about. Of course one expects blurring when the camera is panned/titled/waved-about unless you you a very short exposure time (small shutter angle). Try the same thing with your still camera (image stabilisation ‘off’)with the same shutter speed.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Blog at WordPress.com.

Up ↑