Bob's Tests & Comments on the HMC40

There's math, and there's footage; I recommend actually shooting footage and finding out for yourself how it fares. I find the digital zoom kind of pointless, as it's just a magnification of the center of the image, and 5x is crazy and 10x is just useless (unless you're trying to replicate an out-of-focus security camera look or something).

Digital zoom doesn't really give you anything you couldn't do better in post.

However, there is one potential usage for it, I guess; if you really, truly, absolutely needed the absolute longest recording times, and you were definitely only going to SD/web, then you could probably use digital zoom in the 6-megabit HE mode. HE is heavily compressed, but digital zoom would be feeding it only 1/4 as much data, so I bet the overall compression efficiency would be a lot higher...
 
Bob, just wanted to add my voice to the others saying thanks for your work in this. It's nice to see someone take a camera and really play with it and see what exactly it can do rather than just sticking with default settings!
 
I shot some footage with that 6Mbps mode, and it would downscale to SD okay, but I doubt it would if you did 2x digital zoom.

I don't know how that digital zoom works (never tried it), but I would assume there's interpolation involved that is going to yield what still looks essentially like almost as much detail as an optically zoomed image would to the codec. I've done some experimenting, and compressing video that has been upscaled does not work as well as simply compressing the original video (at the same bitrate) and then upscaling. I was trying to see if I could get similar compression/image quality efficiency at low bitrates as I could using anamorphic video - didn't work very well at all.
 
Btw, that 6Mbps mode actually impressed me. I used it when I first got the cam and only had an 8GB card, but needed to record a couple hours uninterrupted. I was expecting it to really look like crap, but it was actually quite viewable. Seeing that convinced me that the AVC encoder in the HMC40 is a pretty decent one.
 
Ullanta recommends using 1/30 shutter speed with the 6mbps mode, which creates a "fake 30p" look that compresses even better, making the 6mbps mode a more practical choice for longform work.
 
What would have been really nice, is if Panasonic had offered the option of using any of the bitrates with any of the frame size/rate formats.
 
Barry_Green said:
Ullanta recommends using 1/30 shutter speed with the 6mbps mode, which creates a "fake 30p" look that compresses even better, making the 6mbps mode a more practical choice for longform work.
Click to expand...

I wonder how much that would really help compression. It will reduce noise, which can help quite a bit (essentially a crude temporal noise filter), but the camcorder will still be encoding fields separately rather than full frames.
 
subtleinspiration said:
Bob, just wanted to add my voice to the others saying thanks for your work in this. It's nice to see someone take a camera and really play with it and see what exactly it can do rather than just sticking with default settings!
Click to expand...

Thanks for you kind comments.

Whenever I get any equipment, I like to take it through its paces to see how it functions as well as see how far I can push it. That way at a later date, I can know if I can do something reasonably well or not.

The HMC40 is a handful, with so many features and options, it's hard to try everything out.

Bob Diaz
 
Well it seems that between all the rain and getting ready for the start of classes, I haven't had a chance to do a whole lot with the camera. Today, I decided to take the bull by the horns and put other things to the side...

DRS: Very impressive, I'll post the results later on, but I gave it a super high contrast (there's no way it can get everything) shot and it worked really well.

Now Adam Wilt did a rather extensive review of the HMC40:

http://provideocoalition.com/index....w_panasonic_ag_hmc40_1_4_3_cmos_hd_camcorder/

I want to point out on page 3 of the review, he points out an important limit to this technology:

The camera’s Dynamic Range Stretch (DRS) function works like the “Highlights and Shadows” functions in various image editors: it selectively boosts shadows and brings down highlights in localized areas, giving you some of the benefits of high-dynamic-range imaging. Used judiciously, DRS lets you see farther into highlights and deeper into shadows; used to excess or on the wrong scenes, and you may see some “halo” effects:

P40DRS.jpg

Dynamic Range Stretch on high, when the operator wasn’t paying attention.
Click to expand...

DRS is not a cure all for every shooting situation, but used with care it can extend the dynamic range of the image sensor. Look closely around the hat on the person and you'll spot a difference in the brightness in the sky; almost like an outline.


TIME LAPSE:

Today is an ideal day for playing with this feature, we have large puffy clouds that beg to be videoed in time lapse. My back yard faces north, so I can safely set up the camera and go inside. However, for the south facing shots, I'll be sitting on the front porch watching the camera. (I do live in LA, so yes, I really do need to watch the camera.) :)

If all goes well today, I hope that the sunset looks great..... we'll see.

I'd say more, but now that I've gotten started on shooting this stuff, I really want to play around some more...


Bob Diaz
 
This is a very common problem in astronomical imaging where you are often having to deal with a tremendous dynamic range against a limited way to present it. And those halos are the bane of that processing.

DRS used in post for still images (using something like a deconvolution routine or digital development processing) is a destructive technique, which means it changes the makeup of the data itself, as opposed to, say, shifting, scaling, or filtering the histogram, which can still leave the underlying data recoverable. Destructive isn't a pejorative here, it just describes a characteristic of the function. Often, when doing post processing on astronomical images, these types of routines are done in the final steps to: 1) better preserve the data, and 2) reduce the potential of halo artifacts like Bob described above.

So, DRS being used on the front end can be a bit risky. I would use it only when absolutely necessary, because it will be extremely difficult, if not impossible to take out any artifacts in post. Plus, you will be more limited in post as to the extent that you can use highlights/shadows, exposure, and contrast adjustments. Use it when there is a huge dynamic range in your scene that is otherwise overwhelming. If that is not the case, look for some alternatives.


Time Lapse:

I have used this a few times at our astronomy campus lately and this is a terrific feature! Clouds, shadows, flowers, assembly of structures, observatory dome movements, etc. take on a new dynamic. Way cool!
 
OK, here's the test shots with DRS, you'll notice that the file name appears in the image.

00000.MTS & 00004.MTS are DRS = Off
00001.MTS is DRS = 1
00002.MTS is DRS = 2
00003.MTS is DRS = 3

DRS_Off.png
DRS_1.png


DRS_2.png
DRS_3.png


DRS_OFF-2.png


I put in a second DRS = Off, just to compare the background. The clouds were moving quickly and I wasn't sure if the change in the clouds had an impact on the sky part of the image.


Bob Diaz
 
Nice comparisons, Bob.

It would be really interesting to compare the histograms of those images to see what DRS is, in fact, stretching. It looks like the dark end of the scale and the red histogram are being pushed, but I could be wrong. What should be happening is that the histogram distribution curve is refitted to the existing dynamic range, thus bringing up the distribution in the dimmer areas closer to the brighter areas and the brighter closer to the dimmer. In a fully stretch version of this function, the histogram would be nearly flat (and then you get the halos!).
 
One thing you can't see clearly in the still photos is the dynamic impact of image noise in the darker areas of the image. Not that the noise was really that strong.

For this shot, 3 had the strongest noise; no surprise, because it's the maximum setting. While 1 had the lease noise, it just didn't seem to do very much to help lighten the dark areas. Looking at my TV, I'd say that 2 was the best overall choice between keeping the noise small and offering to extend the range.

Keep in mind that this was just one test and other shots may yield different results.


Bob Diaz
 
Cranky said:
If focus ring is not a must, then for this money I would prefer a used/refurbished Canon HF100/HF10/HF11. 17 or 24 Mbit/s depending on a model (Pana: 17 Mbit/s only), 24p and 30p (Pana: 24p only, both in 60i), larger sensor (1/3.2-in vs. 1/4-in), built-in ND filter soaks up to 4 stops of light with 1/4-stop increments, at least as good or better in low light.
Click to expand...

My current choice for a small (four of these take up less space than the HMC40) unmanned camera is the Sanyo Xacti FH-1. I bought one of these ($400) mainly to offload wear and tear on my Sony HVR-A1 (predecessor to my HMC40) when shooting local High School sports. Well, that, and it offers a 1080/60p mode (MP4/AVC at 24Mb/s average)... really nice, versus 1080/60i. And not as bad as I expected on compression artifacts -- AVC has some a long way in the last few years.

I was rather surprised by this one. The single sensor is 1/2.5", and does perform better in low light than my Sony did, painful as that realization was. It also has full manual control: ISO, shutter speed, and aperture (uses still camera ISO settings, rather than the usual gain numbers).

Not that you'd get other features comparable to the better consumer Canons or Pannys, and the audio I've only used for sync. It skips between 4GB boundries when writing to SDHC, and while it does 1080/60p/60i/30p, 720/40p, and 480/30p, there's no independent bitrate control. No optical image stabilization either. But for quality per dollar, it's way up there.
 
You must log in or register to reply here.
Back
Top