Yes, thanks, and I am most impressed with it. After 60 years of looking at NTSC SD it's as if someone has turned the lights on!

Ok A.J., you tipped your hand...at last there is another DVinfo'er with an H1. CONGRATULATIONS!

And since you're technically far more knowledgeable than I am, GET YOUR BUTT TO WORK and do all kinds of torture tests right away! ;-) If you're able to verify the favorable results of my crude tests, that would be wonderful, and I'm sure that you can do a lot better and more detailed job of it.

Sorry, folks, I've been busy this week with things far less entertaining than playing with my new H1, but should have more time this weekend. Top two on my list are pushing to see what it takes to break the inter-frame (GOP) compression (always a contentious topic when people are talking theory but don't have facts/tests), and simply to get some real-world, non-test clips posted for other folks enjoyment/amusement. A.J., anything you can post will be greatly appreciated.

Cheers and again, congrats!

I wonder how efficient the OIS, Optical Image Stabilizer is?
Is it of good advantage if you like to use the camera without tripod?
Is it definitely a feature to have that you should not choose another camera without OIS? I appreciate it in a still camera like my Canon D60 but how good is it for video with panning camera etc?

Top two on my list are pushing to see what it takes to break the inter-frame (GOP) compression (always a contentious topic when people are talking theory but don't have facts/tests), and simply to get some real-world, non-test clips posted for other folks enjoyment/amusement.

Pete/AJ: don't forget to test and post examples of the lowlight capabilities.

Regarding breaking the compression, try mounting the camera on a tripod/dolly setup and rolling it across a rough surface. That's the kind of movement which causes trouble on the Sony HDV cameras.

I'll comment at this point that one of the most impressive things for me about the XL-H1 is that the codec seems to work so well. I did not pick I frames to post because I have no way to tell whether a chosen frame is an I frame, P frame or B frame. Stepping through captured (FCP5) video frame by frame gives no clue as to which frames are which. I tried to "break" the system by putting the camera on a tripod and turning it through 360 degrees as fast as I could. The frames naturally look blurry but the blur appears to be no different from that which you would normally associate with the blur from the shutter speed.

A.J. From what I've seen in my own experience and other people's comments, where HDV suffers problems is if the camera experiences a strong vibration or sudden bump. I particularly notice this when my FX1 gets jarred up and down, but a sideways knock should do the same thing. Would be interesting to know if you can replicate this on the XL-H1.

P.S. Best solution is reportedly to insulate the camera from shock and vibration by any means available, like foam or a beanbag or something.

Here are 4 starting at http://www.pbase.com/agamid/image/53530487. All 1080i HDV frame grabs captured to FCP5 and exported via Quick time to JPEGs. Re-sampled to 1920 x 1080 in Photo Shop.

Thank you very much. Beautiful stuff!

I'll comment at this point that one of the most impressive things for me about the XL-H1 is that the codec seems to work so well.
It certainly does. At the DV Expo booth, they showed some quite-challenging scenes, rippling water scenes etc., which would have given the Sony some serious challenges, but the Canon showed no ill effects at all. I don't know if it performs better in 60i, but in 24F it certainly seems to be well optimized -- the fact that it's encoding 20% fewer frames, plus that each frame is about 20% softer in detail, plus that the frames get encoded using the more-efficient progressive method rather than the less-efficient interlace method, all add up to the 24F HDV encoding looking very solid.

Yes, thanks, and I am most impressed with it. After 60 years of looking at NTSC SD it's as if someone has turned the lights on!

That's the same feeling I got after getting an HDTV and watching some of the stuff on PBS and HDNET back in 2002.

-gb-

Hey Pete,

Many thanks for putting up the rez chart tests - I found them very informative. The notion that the older SD glass would work fine confirms my earlier suspicion when I was asking for a body only version - maybe that will come later as the demand grows.

What I did find most interesting is that the blue and red ghosting ( which was noted on some early tests from France) seems to be present for both the SD and HD glass - however there seems to be none of it present when the SD glass is fitted to the XL2. This would suggest that the glass is not at fault but the problem lies elsewhere.

Looking in Photoshop at just the blue and red channels independently, the ghosting is confirmed as it is quite clearly burnt into the image; the green channel is perfect. It is difficult to fathom but I suspect some low incidence light is bouncing around between some flat surfaces before it gets to the CCDs.

I wonder if other owners can confirm this blue/red ghosting? Also would it be possible to repeat a test at maximum aperture to see if it is any worse?

The comments about chromatic aberration are really bugging me because while I do see the fringing in Pete's images I don't see them in the resolution chart frames I grabbed. There are three of these starting at http://www.pbase.com/agamid/image/53611832 with one each for the 20X HD lens in 60i and 30p modes and one for the 3x in 30f. These confirm that the horizontal resolution of the XL-H1 is about 720 lines though the lens appears to show modulation out to 880 lines (the MC bars are modulated out to swatch 11 (880 lines) but aliasing is apparent in swatces 10 and 11. They also show that there is a little loss in resolution in going to 720p but not much and that the 3x lens appears nearly as sharp as the others (against the chart at least - in the real world it looks softer).

Now I do not see chromatic aberration in any of these three. Color errors yes but chromatic aberration no. My reasoning is that were chromatic aberration present the red and blue images would be larger or smaller than the green image and you would therefore see a change in image size in switching between the R,G and B images. I don't see such a change. To look for it I concentrated on the white ring which bounds the inner portion of the chart. I don't see any color fringing around this ring either. Does anyone else?

++++. Does anyone else?


I would agree with you that your tests in addition to Pete's tests show little Chromatic Aberration.

The blue and red ghosting that I saw in Pete's tests also manefest in your tests; to a different degree but then again the chart and exposure settings are different.

While the images from the H1 are generally good I am trying to discover the conditions which cause the onset of the blue and red ghosts. At first I thought it an issue related to high contrast but a test performed at a french site goes against this theory cf

http://s56.yousendit.com/d.aspx?id=01QIK8OYBGW2H286MOV9TSB2LQ

here you can see a very strong blue ghost down the left edge of the tower and elsewhere

I wonder whether the flat glass window covering the CCD port has a different coating in the XLHI compared to the XL2 (different reflective colour perhaps?). For sure; some light is bouncing around in there.

If there is an explanation, it will be useful to know so that the onset of ghosting can be avoided

It seemed the fairly popular concensus was that the 3x lens was soft until now. I always had trouble with the lens because I found it difficult to check focus with the meager zoom ratio of 3x. Don't get me wrong it's a great lens, and in tight quarters there's nothing I'd rather use except for maybe the mythical 3x manual, but whan you're trying to get focus in a hurry it can be a problem. My question for you is does the HL H1's focus helper function work with the 3x and others or just the stock HD20x lens. Thanks, Jason.

Is it possible that in making the camera PAL and NTSC, that it is mearly a DIP switch that toggles inside the camera housing that is tucked away in a nifty spot? Of course warranty issues would apply, but it's makes you wonder...

BTW I took my Canon L1 apart a couple years back to look at a drum/head issue....it's still in pieces.... So don't try this at home folks!
Now that was a nice lense to work with! (L1) Rack focusing was easier.

larry h.

Is it possible that in making the camera PAL and NTSC, that it is mearly a DIP switch that toggles inside the camera housing that is tucked away in a nifty spot? Of course warranty issues would apply, but it's makes you wonder...

BTW I took my Canon L1 apart a couple years back to look at a drum/head issue....it's still in pieces.... So don't try this at home folks!
Now that was a nice lense to work with! (L1) Rack focusing was easier.

larry h.

In this day and age, dipswitches are becoming rare. Most changes now get done by flashing the eeprom with new firmware. I suspect, but don't know for certain that this is the case with adding the PAL functionality to the H1.

-gb-

In an attempt to get a better handle on what is causing some of these color artifacts I thought I'd take the lens out of the equation and focus (interesting concept with no lens) on the electronics. In SD some color artifacting of the sort I think I'm seeing was created by the fact that color matrixing is done after gamma correction. This causes some crosstalk between luma and chroma signals i.e. changes in brightness in the scene can be reflected as changes in color (which is what I think I'm seeing in my Putora chart samples) and vice versa. The most familiar expression of this phenomenon is the dark band between the green and magenta bars in the SMPTE test pattern. To see if HD exhibits this phenomenon as well I grabbed the bars from the XL-H1. It's posted at http://www.pbase.com/image/53638038. Gross examination of it is sufficient to show that the green-magenta thing is there but blowing it up to the point where you can clearly see individual pixels along the borders of bars/patches is also revealing. Besides the gamma/matrix phenomenon we all have to keep in mind that the color information is subsampled 4:1 which means that while a brightness transition ought to be 2-4 pixels wide a color transition could take 8 or more. Keeping this in mind have a look at the bar pattern. Converting it to Lab and looking at the brightness and color channels separately is instructive. See if you think the color artifacts we've been discussing might be caused by the electronics rather than the lens.

In selling electronics with optional features it is common practice to manufacture only one flavor of the device which contains all the features but to disable the features the customer did not order. If the customer later decides he wants those features they can be enabled often by going to a menu and entering an enabling code supplied by the manufacturer. Other schemes involve changing parameters within the device using special software available only to the manufacturer or authorized dealers/repair stations or, as others have suggested, a new firmware or FPGA download. I'd guess that it is the second method (interaction with the existing firmware) that is involved here. I'd be surprised if it were necessary to actually replace any hardware bits but this is all speculation. I remember a network analyzer where one had to remove the cover, set a dipswitch to "maintenance mode" which enabled new menus and then enter a code in the proper menu from the front panel. The fact that the XLH1 has to go back to the factory for this upgrade implies something like that though it may just be that they are being especially careful to keep the software they are using (if that's their method) safely in house and out of the hands of scoundrels like us.

++++. See if you think the color artifacts we've been discussing might be caused by the electronics rather than the lens.


I looked over the case for electronics and I think that the chart shows a small effect but within the bounds of acceptabilty.

The phenomenon is quite clearly radially dependent whereas an electronic induced effect most likely would apply pan-imagewise?

I delved into the issue more deeply and performed a slant edge frequency response on Pete's 60i ISO chart and the result is here

http://s54.yousendit.com/d.aspx?id=2C22FP2B8182Z2GW08M4VICK9D

the double peak on the blue channel pretty much tells its own story - there is a reflective ghost - just look at the blue channel in isolation and you can clearly see it.

Question is - where is it coming from?

My best guesses are - insufficient coating on the CCD port glass; absence of coating between the microlens (which is bonded to the CCD) and the prism

I would completely rule out the lens as cause and would say it is confined to a region between the lens mount and the CCD surface

I found some discussion on blue ghost behavior here but the author is just as baffled as I am :(

http://www.sd3.info/pf828/Sony_828_Purple_Fringe_Analysis.html

Let's hope it's a friendly blue ghost.

lh

protective filter, it's a protective filter problem, i'd put money on it