Fairly comprehensive test of HV20 modes

[joe12south]

http://www.12south.com/video/hv20/modes/index.html

Due to the raging debate as to whether CINE MODE is destroying low contrast detail, I decided to do a controlled test of most of the different modes and image settings.

I've drawn my own conclusions, but I'll let everyone compare the takes an make their own informed judgements.

[Rikki]

Going by histrogram then :

CUSTOM
ALL 0
f2.4, 1/48
+ 8

Is best?

[iotatau]

I decided to do a controlled test of most of the different modes and image settings.

I've drawn my own conclusions, but I'll let everyone compare the takes an make their own informed judgements.

Thanks a lot for taking the time. VERY interesting for me.

First of all, many of the histograms look "choppy", basically all AE modes. "TV custom all -1" looks at least smooth but it is "exposed to the left", similarly "CINE all -1". TV "0 exp. +4" looks quite good, no highlight clipping as in exp. +8 though maybe some missing detail in dark areas. CINE +8 has the best HV20 histogram. The smoothest histogram overall obviously is the DSLR one. (The Nikon D40x has 10 megapixel, by the way.)

Looking at the images yet another surprise. I compared the best histogram candidates, TV "all 0 exp +4" and CINE "all 0 exp +8". While at first look the TV image has more "punch" it also has significantly more artifacts. There are strong artificial bars at square transitions in the top right, and there are halos and compression artifacts around the letters next to the color circle. The CINE sample exhibits less transition artifacts between squares and much better letters with hardly any artifacts. In my first ten hours I only used P and TV modes but I definitely must try CINE now. Strange that I should think about increasing exposure ("all 0 exp +8"), so far I always tried to reduce the risk of highlight clipping by lowering it.

The D40x picture must have been exposed with a high ISO value. I have a D70, at ISO 200 it produces much less noise and the D40x is about three years younger with a better sensor.

[joe12south]

Everyone, please remember that exposure varies dramatically depending on lighting conditions. You can't just assume that +8 is always a good choice.

[Rikki]

Everyone, please remember that exposure varies dramatically depending on lighting conditions. You can't just assume that +8 is always a good choice.

I was thinking that LOL

[iotatau]

You can't just assume that +8 is always a good choice.

If that's an indirect referral to my post then here's again what I wrote: "Strange that I should think about increasing exposure". Of course I didn't have in mind to set exp. to +8 and to stick with it all the time. I had thought that the risk with the HV20 like with most consumer camcorders is to blow highlights and, if anything, exposure should be lowered. The surprise was that in this case CINE mode with exp. +8 yielded the best result. Note that you didn't raise exposure for the Nikon DSLR (at least you didn't mention it) so in direct comparison the HV20 is underexposing relatively. Looks like one of the CINE mode design features is to provide a reasonable highlight protection.

[joe12south]

iotatau,
I wasn't refering to just to your post, but to the combination of yours and Rikki's ... I just wanted to make sure nobody got confused about the issue.

- Make no mistake, it's super easy to blow highlights with the HV20. I very purposely under-exposed the charts for a variety of reasons relating to judging the sharpness, which was my main goal. I included the various exposures at the end as an aside.

- Yes, one of several things CINE does is protect highlights. It also tries to stay at 1/48 shutter, favors the aperture over gain and has a more flat gamma curve. Unfortunately, it's also losing a lot of low contrast detail. :-(

[24Peter]

Joe - You're to be commended on your efforts, but I'm not sure anything conclusive can be drawn from your tests - other than the Nikon D40 have a very noisy sensor and 4.2.0 colorspace of HDV really struggles with the red channel.

While it is preferrable to avoid blown highlights, for the vast majority of us, some over-exposure in an image is perfectly acceptable. In fact, it may be impossible to avoid a little highlight damage in order to correctly expose a scene. The only exception would be if someone was doing a filmout of their footage. But last time I checked prices were around $200-$300/min to transfer your video to film - and that was without the cost of color grading and release prints.

There may also be times when you want that blown highlight "bleach-bypass" look. Point is, artistic sensibilities may be more important than what proper exposure would otherwise be.

And manual exposure settings like +8, +4 and -11 don't really tell us much and can be dangerous if applied indiscriminantly.

Yes it is helpful to know how the various camera settings affect the image relative to each other. But I would tell most people to hook their HV20 up to a television and see for themselves how the image changes (in real time) by running it through the different settings.

[joe12south]

Yes it is helpful to know how the various camera settings affect the image relative to each other.
That WAS the whole point of the tests, especially as it relates to sharpness, and that was accomplished.


But I would tell most people to hook their HV20 up to a television and see for themselves how the image changes (in real time) by running it through the different settings.
Experimenting in the living room under random lighting conditions reveals certain things, comparing charts in a controled setting reveals others. It takes a lot of time to run through 50+ set-ups and this is knowledge I wanted to have.


And manual exposure settings like +8, +4 and -11 don't really tell us much and can be dangerous if applied indiscriminantly.
They tell you A LOT about the results of using gain and low light noise rendition. Again this is knowledge I wanted to have before getting on set. That's exactly why I tested them, so that I would not run into the "danger" of applying them "indiscriminately. "


Nikon D40 have a very noisy sensor
If you concluded that, then you really came to the wrong conclusion!