Pixellation in Strobe Lighting
 

Sigh.

I'm the videographer for a band and I've been filming their gigs lately. Every time I've filmed one, there's been a problem with the camera. First it was the onboard microphone being too sensitive for its own good, and distorting completely, and also honing in on the low to low-mid ranges of the equalizer so cutting out the vocals; second time, I had to use nightshot 'cause the venue was so dark - but what does the camera decide to do? Make it DARKER by increasing the shutter speed from 1/25, as I'd set it, to the 1/50 standard again... and then the last gig, I had this problem;

There was a lot of strobe lighting, which, in some cases did conflict with the shutter speed, but, another thing that happened, was, the image, when I watch it back on a TV or on the camcorder's screen - there's areas of really bad pixellation in the image. When the strobe lighting stops, this stops. Any idea why this might be? Is it anything to do with it being a CMOS instead of CCD? Grr, why can't this camera just be RAW and stop overdigitising everything, broadcast camcorders can manage with strobe lighting so why can't this...

Yeah so is it the CMOS d'you reckon?

You never mentioned what camera and in what mode you are filming: SD DV or HDV MPEG?
I suspect that you are shooting HDV MPEG.
In my opinion, what you are describing are typical symptoms of mpeg (the cam) not being able to encode the fast luma changes in the adjacent frames.
So this has nothing to do with CMOS vs. CCD.

Do you have the camcorder mic levels set to manual rather than automatic? Automatic is basically an automatic gain controller, it often causes problems with quickly changing sound levels. For manual, it's best to set it a little below the point where it would max out on the loudest sounds.

It's the mpeg encoder doing its thing. Pretty bad looking isn't it? I don't think these newer hdv-cams exhibit it as badly as hc1 or do they? I assume you are using the hc1/a1?

yes, i've seen the same thing. it's the compression used in the HDV codec. at this point you realise that HDV is definitly a consumer format, because as soon as you have significant changes from one frame to the others it all breaks down.

the reason for this is that HDV, in order to save space on the tape, only stores the changes from the previous frame. if the changes are too big it can't fit all the information on tape and something has to give, usually resulting in a dramatic decrease of resolution (big blocky areas in the image)

one way to get around it is to switch the camera to DV recording. yes, it's SD but at least every frame is totally independently recorded from the previous one so strobes and fast motion do not affect the quality of the recording.

Ahh wow, thanks, that explains everything. yeah , sony A1 in HDV format. I WONDERED how it managed to fit HDV onto the tape...

Would it not compress if I was using an 'HDV' tape?

Thanks a lot, I'll film any stroboscopic gigs in DVCAM from now on.

Or would you recommend that I just shoot in DVCAM all the time? Is the A1's HD standard not really very high? I mean, you say it's mpeg so just changes the necessary pixels, so surely this would give visible pixellation? 'Cause in any image, every pixel should be changing on every frame, since it's... motion photography. And I don't want any dodgy pixelly mpg encoder effects.

So what would you recommend? Good ol' true DVCAM?

As for the audio, yeah, I could set it manually but the thing is when a song finishes I want the sound of the audience... which means quickly going into the menu and changing it... my PC101 managed fine at loud concerts, the A1 should be able to cope, albeit a more sensitive microphone, but it should be built to deal with that fact.

I'd say stay with HDV unless your having some kind of strob problem.

There are plenty of Pro HD cameras which record as MPEG2. I've not had any problems with artifacts, event during camera flashes at the weddings I've shoot.

A down converted HDV will be better than DV.

As for audio, I allways use manual levels, but I have gotten blown away a couple of times when I wasn't ready for the organ to kick in at some weddings. Some day when I can afford another shootgun I'll use two, each set at differant levels. I'm not holding my breath.

Jonny -
You're going to have SOME compression, even DV/DVCAM is compressed - there's no way to record that much information "uncompressed" without doing something to reduce the size of the information from each frame...

HDV does pretty well, I shot some stuff at a dark party with a strobe going (HC7) - it looked rough, but I was looking to see how bad it really would be... seemed to handle it a bit better than my HC1 which sometimes even had a bit of trouble with a flash. I got footage that in a pinch could be doctored up in post, but I wouldn't want to count on it. Keep in mind that a strobe light can cause our vision to become disoriented, and presumably a human being has slightly more powerful processor <wink>! Technology can only do so much.

Thing is even SD DV cams aren't "perfect" - I went back and looked at some footage off a TRV900, and it had issues when flashes went off too. Different, but distorted nonetheless.

The thing that makes it so much more noticeable with HDV is the clarity of the picture when everything is going as expected... it makes the "flaws" much more obvious...

As for audio, any time you're dealing with high SPL/db it's going to be a challenge - you might want to try using headphones to monitor the audio and ride the levels or set a level that isn't too distorted.

pro HD cameras (i.e. non HDV) record with much higher data rates (i.e. less compressed) which is why they are better.

no that is absolutely not true. if the HDV footage has significant mpeg2 compression artefacts a down convert will be worse than a DV original - since it can't remove those artefacts and they are still very visible at SD resolutions. and there is no way DV compression exhibits those type of artefacts.


HDV and DV tape are essentially the same - so that wont help.

The HDV standard in general is not very high - no matter what HDV camera you use. for slow moving subjects the results tend to be acceptable though and typically much better than DV.

and yes, pixels change from frame to frame. mostly. the larger the change the larger the amount of new information to be encoded. but mpeg2 also uses motion prediction - that is tries to take into account where objects have moved to and compares old pixel values with new ones in a different location. when motion is very large this breaks down.

And still my HC1 and HV20 cameras outshine the dvx100 in the image quality department.