tests of dynamic range with DVRack + comments
 

i put the camera through some tests with dvrack (non-hd) last week. watched the histogram through various modes and settings, locking exposure when needed for comparisons. some conclusions:

(1) cinemode definitely gives you a more linear exposure result, and it looks like more useable dynamic range. highlights don't clip as early; there are more detail in blacks. unlike the experience of someone in another thread: i see blacks at 0.
(2) the "custom" effects setting gives you more dynamic range if you lower the contrast to -1 in cinemode and tv modes (the ones that i tested). this is unlike the contrast control in, say, the panasonic gs400 which simply lowers the output range!
(3) it looks like the "brightness" in custom effects controls the auto-exposure... i set this to -1 because i prefer to keep the exposure out of the highlights as much as possible and postprocess for a less abrasive treatment of highlights, trading off, of course, more noise.
(4) this is a nasty one: when i perform a camera tilt (up and down motion) in cinemode i get noticeable transient image warping (a frame is squashed about 3% horizontally--it's skinnier for a frame or two) . when i get time i'll post an example. i can live with this though. it is an amazing camera.

some workflow comments:

- as noted in many posts: sony vegas doesn't yet support pull-down removal for this camera because of the missing flag in the mpeg stream and vegas's inability to force pull-down removal when it doesn't automatically detect it... so unless you do something else, you'll be looking at a gross 3-2 cadence.
- after reading one of david newman's posts about the hv20 and pull-down removal, the somethign else i decided to do was buy connect hd (had used the trial version a long time ago for a multi-cam hd project and was impressed). if you push the pull-down removal button you'll get beautiful 24p output that plays full-speed even on my not-latest hardware. (as an aside, i just upgraded to windows media player v11, before that with v10 it didn't quite play full-screen and full-speed).
- i use connect hd to capture footage and convert directly into CFHD without leaving behind an mt2s. it works pretty well although sometimes i get hangs during the capture (i need to file a ticket for that). it also looks like connect hd stops capturing and goes into convert mode whenever the total file size in the TEMP directory that it creates during capture equals 4gb, using in its calculations even leftover files from hangs (need to file a ticket for that too).

other observations:

- slight barrel distortion when full wide angle
- no noticeable chromatic abberation on wide (not like the xh-a/g1 with it's red-green separation)
- noise in lowlight is much more luma-oriented and agreeable than other cameras i've used (hvx200 for example).
- you couldn't bring this thing into a professional shoot without a bunch of black paint and a mattbox and important-looking things connected to it (and maybe lots of mumbling of "state of the art" and "hot technology")
- killer combination when paired with a merlin
- i have 24p slowmo footage from 60i-->60p that looks very very similar to almost identical footage i shot with the hvx200. remind me to post that comparison too. this camera is a ridiculous bargain.

Thanks for this update. Very useful info.

Thanks for this Ali

Can you confirm what modifications (if any) can be made when using the cine preset? Can you adjust the sharpness for example?

Yes thanks Ali. Good info. Hoping a Vegas update is released in the next few weeks to handle pulldown removal for Sony V1U 24p files. The consensus seems to be that will work for 24p files from the HV20 too.

Thanks for the very in depth input on the HV20. I'll be looking forward to seeing the test footage.

Lots of good info, Ali. I'm particularly interested in your observation that:



You may be referring to my comments. My first observation of this was in comparing two clips (here: http://www.dvinfo.net/conf/showpost....3&postcount=23) by eye. Because I was shocked that the blacks in the cinemode clip didn't look quite black enough, I pulled a still from that shot and looked at it in Photohop which is where I found 'proof' (if you want to call it that) that there were no real blacks. I looked at both the histogram and also used the eyedropper to get info on the very darkest places in the frame and found no pixels that were completely 0.

There are areas of that frame which are looking into the depths of darkness as far as I can tell, which is why I figured there'd be some kind of pure blackness. The comparisons that Chris posted did show blacks, but he had also tweaked the exposure down so I'm presuming that was the reason.

Can you post a clip of cinemode without any exposure or contrast adjustments? I'd be interested to see since, so far the stuff I've seen in un-tweaked cinemode looks to me like the blacks aren't quite there.

Thanks again for your post. Great info.

I've seen this too - any quick movement really, pans or tilts, and the image lags and distorts a bit. I had a wide-angle lens on, and it made it even more apparent. Just one of those things that it's good to be aware of and work around. It's a CMOS thing. (For instance, we all know that low light happens, and wind while outside makes it hard to hear someone talking in an interview - no biggie - we are aware and work around those facts of life.)

I agree.. but, I'm going to show up with it. Most likely with a Z1U in tow as well, but, I won't lie - I prefer the Canon's image. I'm dressing mine up with Carbon Fiber rails, a matte box, filters, an external 8" LCD screen and a XLR shotgun mic, 50' component cables, etc, so it'll be hidden in the sense that an average joe won't know where the actual camera ends and the fluff/extra candy begins. Where I can pull it off, the Z1U will most likely be a fake A camera, and instead play a B camera roll. Yes, I'm serious.

I'd love to see that!

Make sure you post up a photo as soon as you add all these bits and pieces!

Very nice and encouraging report. Any guessitimate on latitude vs an HVX200?

thanks for all the comments. i haven't compared the hvx dynamic range to the hv20. it wouldn't be fair for me to make any guesses about it off the top of my head. the default hv20 gamma curve looks like it's the standard video curved-cliff, and it's hard to extract details from highlights. the cine-gamma mode looks much more linear. i'm sure at some point i or someone else will compare them directly. sorry!

yes there is a separate "custom" effects setting that lets you control: color, contrast, sharpness, brightness (which looks like exposure comp for the auto exp). this custom effects setting is active through cine-mode. unfortunately each setting can only vary through -1, 0 or 1. pretty crude. but better than nothing.

You mean if you want to adjust color, contrast, sharpness and brightness you can only choose between -1, 0 and 1?! I thought it would be more like from 0 to 20.

Thanks Ali - do you know if the -1 compared with +1 setting for sharpness affects the image noticably?

The jumps between -1, 0 and +1 are pretty noticeable and bigger than you'd think. In "EXP" mode (which is essentially manual exposure on the HV20) the range is usually -11>>+11.

Yes, as I mentioned in the previous post, changing the custom settings does affect the image pretty noticeably. Enough that right now I'm leaving sharpness and contrast set to 0, color depth is +1 for a more saturated look.