Hi there,

if anyone is interested, I uploaded some HV20 still images to Mediafire. The images are captures from an uncompressed video stream (8-bit 1080i59) recorded from the analogue component output of the camera, bypassing the HDV compression. The same shots taken from a HC1 are also included for comparision.

There are some minor problems, I encountered when playing around with the camera.

(1) HDMI port does not work with currently available capture hardware.
When connecting to a BM intensity card, the HDMI output falls back to DVI
mode and scales down to 480p. A HD-Connect MI converter maintains the correct resolution but seems to output the wrong colorspace (RGB instead of YCrCb).

(2) When zooming in aufofocus mode, the camera gets out of focus
and then becomes sharp again continously. This is not a compression side effect, the behaviour is also visible in the uncompressed signal.
Best solution is to switch off aufofocus during zoom recording
(see hv20_zoom1-4.jpg).

(3) When switching to 24p, the image becomes brighter and it seems that also the gamma is affected. As expected there is also an increase in SN ratio. When returning back to 60i the brightness remains at the same level. There seems to be no loss in resolution between 60i and 24p modes.

(4) I miss a color bar function. I have included a 1920*1080 jpeg image which can be put on the camera SD card to ease calibration of a monitor during playback (img_0101.jpg).

Link to the zip archive: www.mediafire.com/?ad3yjrjjwjg File: hv20caps.zip

Regards,

Heinz Bihlmeir

Thank you for your input Heinz. Your findings and contributions are important enough that you should consider starting a new thread within the "Canon HV10 / HV20 HDV Camcorders" area with the same imformation you just posted here. That way people will be on topic when they are interested in the information you've outlined - which I think is important information!

Cheers!

Hi there,
(1) HDMI port does not work with currently available capture hardware.
When connecting to a BM intensity card, the HDMI output falls back to DVI
mode and scales down to 480p. A HD-Connect MI converter maintains the correct resolution but seems to output the wrong colorspace (RGB instead of YCrCb).


This news really bums me out. My major excitment over this camera was to capture uncompressed 24p to a micro-atx pc. I wonder if the issue will be able to be overcome with new Intensity drivers?

It looks like they released Intensity 1.3 today. Can anyone confirm that 1.3 does/does not work with the HV20?

I wouldn't be too worried about the Intensity drivers right now. I'm sure it is hard for Blackmagic to updated their product before the camera actually comes out. Now that some cameras are shipping I'm sure Intensity will be updated. I mean it makes no sense for them not to add support for the HV20 since this card is designed to work with HDMI ports on cameras and the HV20 is one of the few cameras to actually have this. The more cameras the card supports the more cards will get sold so I am sure support is right around the corner.

I'm not sure on this guys. But does this mean the footage that comes out of the HDMI port is prior to HDV MPEG2 encoding? Is it also prior to down rezing to HDV spec as well?

If someone could clarify that be great (if indeed someone has had the chance to test their cam yet).

Thanks for your efforts guys. Can't wait for the Aussie PAL one. DAMN IT HURRY UP CANON.AU. I'm getting a PS3 tomorrow and i really wanted to capture the HD moment in HD. lol :) no chance of that though. Not even listed on their site yet. *cries*

The signal the comes out of the HDMI port at the time of filming has to be uncompressed to avoid the delay caused by the HDV compression process. However, once the signal is recorded to tape and played back through the HDMI you are going to be getting a compressed signal because that is what is stored on the tape. Because the CMOS chip is full 1920x1080 you are also getting full resolution stream, until it goes to tape of course where it is scaled down for HDV.

Rob, i understand the processes etc of it. I was just wanting clarification specifically on the HV20 now some guys have them. i.e. are we getting the stream before full HDV compression (MPEG2 and pixel) and/or before pixel down sizing as well. Or is the HDMI port simply for playback of whats on tape. etc :)

Cheers

There is no HDV compression and no delay when capturing scenes with the HV20 using component or HDMI outputs. If the negotiation with the target capture device is correct, the HDMI output provides 8-bit RGB with 1920*1080i resolution. If this is the true 1920*1080 image from the sensor (after the initial DSP stage) or an upsampled version of the scaled down HDV image is not so easy to determine. There remains also the question, is the final output sampled in 4:2:2 or 4:2:0. Analysis of the captured frames may answer these questions.

Cheers

Your findings and contributions are important enough that you should consider starting a new thread within the "Canon HV10 / HV20 HDV Camcorders" area with the same imformation you just posted here. That way people will be on topic when they are interested in the information you've outlined - which I think is important information!Agreed, therefore I have split these posts out from another discussion so that they stand as their own thread.

In the "HV20 1440x1080 or 1920x1080?" thread, I officially offer beers and hosting of the raw files for anyone wishing to post a res chart test of HDV vs Component vs HDMI. Or just HDV vs Component or HDV vs HDMI. Bonus beer if you test 24p vs 60i too...
http://www.dvinfo.net/conf/showthread.php?t=89603

Bruce Allen
www.boacinema.com

Hi Heinz,

thanks for this, also from me. I'm fascinated by your point (3) and the two test result JPEGs. How do you explain the higher noise levels in 1080i?

Thanks,
Pieter

oh, and any thoughts why the cyan falls short of the other gamuth extremes? Lighting or camera?
tnx

There is no HDV compression and no delay when capturing scenes with the HV20 using component or HDMI outputs. If the negotiation with the target capture device is correct, the HDMI output provides 8-bit RGB with 1920*1080i resolution. If this is the true 1920*1080 image from the sensor (after the initial DSP stage) or an upsampled version of the scaled down HDV image is not so easy to determine. There remains also the question, is the final output sampled in 4:2:2 or 4:2:0. Analysis of the captured frames may answer these questions.

Cheers

I'm a little confused, you're stating there's no HDV compression out the hdmi port and then saying you're not sure what output res and color sampling is being output. Does anyone know for sure here?

@Pieter

I assume its the lighting. The noise level may be affected by the fact, that in interlaced modes two independent fields are captured at different times. Also, the camera was a NTSC version, and there may be some modulation caused by interference with 50Hz light sources (here in Europe).

@Rich

The problem is, that the output of the single chip CMOS sensor (with bayer filter) is processed and enhanced by a DSP to generate the YCbCr output signal. The signal may be downscaled during this process or an additional resampling stage may be used. The HDMI interface provides an independent scaler to support different monitor resolutions.

I don't think that the real output resolution captured from the HDMI output is better than HDV. There are of course compression side effects, depending of image content and motion in the scene.

I have captured and analyzed a resolution chart using HDMI and HDV and compared the results, they are quite simular (see attachment).

The optical system (lens, focus etc.) is also limiting the bandwidth.

Regards
Heinz

There remains also the question, is the final output sampled in 4:2:2 or 4:2:0. Analysis of the captured frames may answer these questions.

If it's 4:2:0, and captured with component, isn't there a chance there will be lag on the chroma side and we might see a ghost trail or such (just of the chroma?) That might be a giveaway. Not sure.

Anyone have ideas how to tell 4:2:2 from 4:2:0 on captured footage? 4:2:0 should be "half" of 4:2:2 in some regard, right?

(2) When zooming in aufofocus mode, the camera gets out of focus
and then becomes sharp again continously.

This sounds like the same issue addressed in another thread titled "Returning HV20 After One Hour." After several pages of discussion, a consensus seemed to develop that the original post was based on unrealistic expectations. But Heinz Bihlmeir has made a similar observation.

Now I'm really confused. Does the HV20 focus-hunt during zooming more than other camcorders? Is it a defect, or should one just accept that autofocus must be turned off during zooming? More feedback on this subject would be appreciated.

Mike, The default instant auto focus us just that, way fast, in this case calling it instant isn't much of a stretch. If the center of the image is small, like a squirrel, it will hunt if the camera person is shaky with the camera or the squirrel moves, not so with a larger subject or a steady hand. It is a very useful auto foucs feature, you can even simulate a rack focus with it, it snaps to that fast, it is unlike any other auto focus I've used and a great tool. I did change mine to regular ( not instant ) in the menu and it works like any consumer auto focus. I've been using that setting, but when going with manual focus I set back to the instant mode and can go to that or manual quickly that way without the menu. To each his own, but there are no issues with HV20 focus. None at all, now like any auto focus when you get in low light performance of the auto focus degrades, I'd suggest going manual and using the focus assist on the LCD when lighting dims. The HV20 image does do well in lower light, at least in 24p mode, which took me by surprise. I'm in no way saying it is a night vision device, you need light, but it does well in that regard for a basic consumer camera. My plan is to use an external monitor for narrative work, but that is due to the addition of the Brevis35 on front and a Nikon 50mm. In that configuration I will focus using a 720p monitor, not needed for the camera on its own though.
Thanks, Alan

..just waiting for my Brevis35.

Does the HV20 focus-hunt during zooming more than other camcorders? Is it a defect, or should one just accept that autofocus must be turned off during zooming? More feedback on this subject would be appreciated.

Here there,

maybe its a fault of my Japanese HV20 with an early firmware revision , but I can easy replicate the described effect. If anyone is interested I can upload a video made in auto mode, which shows the effect. It doesn't disqualify the quality of the auto focus system in general, but there seems to be a certain condition (lighting, scene content, distance etc) in which the closed loop of the AF system becomes slightly instable. The four images in the attachement are consecutive frames (24p) from a record during zoom.

Heinz

PS: I love the camera and I can live with this problem.

This sounds like the same issue addressed in another thread titled "Returning HV20 After One Hour." After several pages of discussion, a consensus seemed to develop that the original post was based on unrealistic expectations. But Heinz Bihlmeir has made a similar observation.

Now I'm really confused. Does the HV20 focus-hunt during zooming more than other camcorders? Is it a defect, or should one just accept that autofocus must be turned off during zooming? More feedback on this subject would be appreciated.

No Mike, in fact it hunts far far less than any other camcorder I've ever used...sans the HV10. There appear to be only two reports of this with the HV20 and everyone else is having no issues. I don't know why people will always focus on 1 or 2 reports when there are so many others that report no issues. We don't know if 1 or 2 cameras were defective, if the user was unaware of how to use the camera or some other issue. But the bottom line is if this is such a tiny % of the total users, it's obviously not a design issue.

Heinz, I would suspect it's some aspect of an early firmware version. It seems that you and only one other person reported this issue. Everyone else is having no problems. I compared mine with my HV10 quite extensively and it simply never missed when zooming at full zoom or full wide angle. It was every bit as superb as the HV10.

Hi Ken,

there is another. ;)

Joe Busch also reported it:
http://www.dvinfo.net/conf/showthread.php?p=647508#post647508

Now i'm sure this is something I will be able to live with (as in be annoyed with when it happens but getting over it because of the rest of the package is so good).

If I'm not mistaken my (currently dead) Optura 100MC also had trouble focussing while zooming. I solved that by zooming in first, locking focus, then zooming out for the beginning of the shot, and zooming in slowly during the shot while keeping focus locked.

I'm unclear how the HV20 w/ identical optics/OIS would have focus problems in zoom, that the HV10 didn't (apparently) have.?
Shouldn't the camera stay in focus all during a zoom in/out, once focus is set? (old trick to focus at full telephoto, then zoom out for wide w/ focus locked, so when you go to closeup during the shot you don't get hunting)

The only issue I have w/ zoom & focus on the HV10 is that at far telephoto for relatively near (5-10') objects, it goes (and stays) out of focus (but as tipped here, I should stay wide and use macro-style instead).

Hi Ken,

there is another. ;)

Joe Busch also reported it:
http://www.dvinfo.net/conf/showthread.php?p=647508#post647508

Now i'm sure this is something I will be able to live with (as in be annoyed with when it happens but getting over it because of the rest of the package is so good).

If I'm not mistaken my (currently dead) Optura 100MC also had trouble focussing while zooming. I solved that by zooming in first, locking focus, then zooming out for the beginning of the shot, and zooming in slowly during the shot while keeping focus locked.

Pieter, the original thread starter has since exchanged her first HV20 for another unit and she reports it works perfectly. She says the zoom & focus are now perfect. So that one was unquestionably defective. I still think this represents a tiny % of the total units out there. But we can say with 100% certainty this is not a design issue. If it were, virtually all units would exhibit the same behavior. Also keep in mind that the design of the Instant Autofocus is entirely different than what was used in the Optura.

Colin, you are correct, there is NO difference in the design of the Instant Autofocus in the HV10 & HV20. They both work perfectly, but there may have been a couple of problematic units. I see absolutely no difference whatsoever between the Instant Autofocus during zooming of my HV10 & HV20. They both work great.

The procedure you described is a manual focus procedure, but is generally unnecessary if you want to use autofocus. An exception might be during low light, shooting through glass or something unusual in the field of view.

Hi Ken,

Pieter, the original thread starter has since exchanged her first HV20 for another unit and she reports it works perfectly. She says the zoom & focus are now perfect. ...

Great, didn't catch that post. That /is/ good news.
Thanks,
Pieter

I tried out an HV20 in a store yesterday and actually taped some of the footage and took it home....

The one I tried definitely had focus challenges when zooming. Especially when the scene had more than one object in it. It basically was not useable footage when zooming. That's what I found anyway. It did work for simple focus where the object it is focusing on is the only thing in the picture. That being said, it's dirt cheap and something that can learn to be lived with....

I tried out an HV20 in a store yesterday and actually taped some of the footage and took it home....

The one I tried definitely had focus challenges when zooming. Especially when the scene had more than one object in it. It basically was not useable footage when zooming. That's what I found anyway. It did work for simple focus where the object it is focusing on is the only thing in the picture. That being said, it's dirt cheap and something that can learn to be lived with....


Mike, it's quite possible that there was an issue with that unit. I can tell you that most people and reviews are finding this to be the best autofocus in any camcorder. The original poster thought there was a problem too until she got her new one. She now raves about it. :)

I tried out an HV20 in a store yesterday and actually taped some of the footage and took it home....

The one I tried definitely had focus challenges when zooming. Especially when the scene had more than one object in it. It basically was not useable footage when zooming. That's what I found anyway. It did work for simple focus where the object it is focusing on is the only thing in the picture. That being said, it's dirt cheap and something that can learn to be lived with....What store in Canada did you find it in??

Please get back!

Mike

thanks for doing that test! it makes sense since, in a static scene, only the color resolution is different between hdv and hdmi, correct?


@Pieter

I don't think that the real output resolution captured from the HDMI output is better than HDV. There are of course compression side effects, depending of image content and motion in the scene.

I have captured and analyzed a resolution chart using HDMI and HDV and compared the results, they are quite simular (see attachment).

The optical system (lens, focus etc.) is also limiting the bandwidth.

Regards
Heinz

thanks for doing that test! it makes sense since, in a static scene, only the color resolution is different between hdv and hdmi, correct?

Yes, the vertical chroma resolution is better when the signal is captured via hdmi (hdmi:422, hdv:420).
If the handshake with the hdmi display succeeds, the 1440*1080 420 image is upconverted to a 1920*1080i 444 YCrCb format, else the image is downconverted to 480p RGB (DVI).

can you clarify...

is the live signal coming out of the HDMI port is 1440x1080 4:2:2 upscaled to 1920x1080 4:2:2?

so it isn't true 1920x1080?

for problem #1 below, i filed a ticket with blackmagic and sent them a link to this thread. they responded and said they will check this out. thanks heinz for the testing!


Hi there,

if anyone is interested, I uploaded some HV20 still images to Mediafire. The images are captures from an uncompressed video stream (8-bit 1080i59) recorded from the analogue component output of the camera, bypassing the HDV compression. The same shots taken from a HC1 are also included for comparision.

There are some minor problems, I encountered when playing around with the camera.

(1) HDMI port does not work with currently available capture hardware.
When connecting to a BM intensity card, the HDMI output falls back to DVI
mode and scales down to 480p. A HD-Connect MI converter maintains the correct resolution but seems to output the wrong colorspace (RGB instead of YCrCb).

(2) When zooming in aufofocus mode, the camera gets out of focus
and then becomes sharp again continously. This is not a compression side effect, the behaviour is also visible in the uncompressed signal.
Best solution is to switch off aufofocus during zoom recording
(see hv20_zoom1-4.jpg).

(3) When switching to 24p, the image becomes brighter and it seems that also the gamma is affected. As expected there is also an increase in SN ratio. When returning back to 60i the brightness remains at the same level. There seems to be no loss in resolution between 60i and 24p modes.

(4) I miss a color bar function. I have included a 1920*1080 jpeg image which can be put on the camera SD card to ease calibration of a monitor during playback (img_0101.jpg).

Link to the zip archive: www.mediafire.com/?ad3yjrjjwjg File: hv20caps.zip

Regards,

Heinz Bihlmeir

can you clarify...

is the live signal coming out of the HDMI port is 1440x1080 4:2:2 upscaled to 1920x1080 4:2:2?

so it isn't true 1920x1080?

No it isn't, this is my conclusion interpreting the results of my measurements (see attachment posted earlier in this thread).

Btw: The HD-Connect MI HDMI->HDSDI converter works now with the HV20 in the correct color space (fw 1.1.35).

Regards,
Heinz

Ken, I have noticed some instant autofocus "peculiararities" on my HV20. The manual specifically recommends turning off IAF if you are shooting through a window pane. I was filming a scene through my living room window the other day with a slow zoom to telephoto, and the trees outide were rapidly pulsing in and out of focus. Yesterday, I was filming elk in a meadow at sunset. While panning rapidly in full telephoto, the trees on the horizon were pulsing rapidly in and out of various degrees of focus.

In one case, I was using IAF in a situation that the manual cleraly states will cause problems. In the other, I was panning so rapidly that the footage would have been unuseable anyway.

In normal use I have found the IAF to be excellent. I wonder if there is some middle ground between the idea that some of these units are defective and the likelihood that there are some situations for which normal autofocus is a better choice. The camera provides you with the option to use either. No setting on the camera will allow it to "read your mind" and discern exactly how you want your footage to look. Even in normal autofocus this camera locks onto focus more accurately and decisively than my HC1. Since IAF is so agressive, perhaps there are simply times when you will want to use normal autofocus instead.

Pat

Pat, virtually all of the conditions you cited are extremely difficult for any autofocus. Shooting through glass is a situation always mentioned in any manufacturer's camcorder instruction manual. You may be better off using manual focus for that. The other condition you mentioned is also very rare and something you should generally avoid for a number of reasons. It's not good practice to pan while at full zoom. First off no viewer wants to watch that kind of video...it's dizzying. Secondly, the autofocus just won't work properly. If your subject distance is the same while you do those full zoom pans, you're far better off using manual focus and leaving it there.

That would be true with any cam. You just have to come to a reasonable expectation of what any of these systems can and can't do. But feel good in the knowledge that the Canon Instant Autofocus is probably the best system out there.

I absolutely agree with everything you just said. My only point is that IAF is something new. Since it can agressively achieve focus in situations that challenge normal autofocus, users may sometimes see it going in and out of focus and think that there is something wrong with the camera, when there is not. (My only reason for recording a rapid pan was that I didn't want to shut the recording off while I tried to catch an elk briefly in view in an area free of distracting foreground clutter.)

Pat