XH A1 compresses in HDV then writes in DV to tape? at DVinfo.net

Roland Oeller · May 7th, 2007, 05:19 AM

Hi!
I have now shot my first project in SD DV (after shooting some stuff in HDV) and while capturing I noticed that the image looked somewhat MPEG2-compressed.

This should not be since I recorded in 4:3 DV 25f. Also I played the tape on an old DV-camcorder so the signal is definitely DV. I captured in Premiere Pro 2 with the usual DV-settings.

I suspect that the A1 internally uses the HDV signal as a basis and converts it to DV. The compression is not that obvious but it is there.

You may have to download the image to be able to zoom in and see what I mean.

http://img143.imageshack.us/img143/1...ession1bm7.png
Here is a portion of the image resized to 400%:
http://img516.imageshack.us/img516/6...ession2pc9.jpg

Look at the lower edge of the pit with high contrast and the edge of the wheelbarrow. There are some obvious compression artifacts there. And look at the blue trousers which show macroblocking.

This is not a huge problem for me since I care more about HDV but I find it disturbing that nobody seems to have noticed this yet. A DV-image should not look like this, right? Maybe it is only something the PAL-Version of the camera does? Maybe it´s the Frame-mode? Or are my eyes playing a trick on me?
Any thoughts?

Mikko Lopponen · May 7th, 2007, 06:02 AM

Hard to say. DV-compression does use mjpeg and it also causes those jpeg-compression artifacts. Shoot something where there's A LOT of motion and it should be apparent if it's gone through the mpeg2 encoder.

Pavel Sedlak · May 7th, 2007, 10:22 AM

I have the same experience, DV have some mpeg2 compression residue.

When you make SD mpeg2 (from this DV), it is more noticable.

Don Palomaki · May 7th, 2007, 10:37 AM

From a RFC:

"The DV format only uses the DCT compression technique within each frame, contrasted with the interframe compression of the MPEG video standards. All video data, including audio and other system data, are managed within the picture frame unit of video."

"The DV video encoding is composed of a three-level hierarchical structure. A picture frame is divided into rectangle- or clipped-rectangle-shaped DCT super blocks. DCT super blocks are divided into 27 rectangle- or square-shaped DCT macro blocks."

So with the DV format, upon close examination, you can expect to see some macro block artifacts. However, because each frame stands alone compression wise, they may not be as apparent in video (as opposed to individual frame grabs).

Bill Busby · May 7th, 2007, 10:50 AM

My take on this. Blow ANY DV signal up to 400% & you'll see the same results.

The original pic you posted looks fine here.

Bill

Pavel Sedlak · May 7th, 2007, 11:57 AM

Blow ANY DV signal up to 400% & you'll see the same results.

http://www.videoproduce.cz/images/DV_400pct_vx2000.jpg

it's not look like the same result (specially about edges)

Bill Pryor · May 7th, 2007, 12:45 PM

How does it look when playing back, not from a freeze frame?

Pete Bauer · May 7th, 2007, 12:47 PM

The sawtooth along high contrast edges in Pavel's picture look more like interlace artifact from a slowly panning camera than compression artifact.

Pavel Sedlak · May 7th, 2007, 02:51 PM

Pete, you are right. Only interlacing, nothing else (like XH-A1 dv).

I mean that difference is too big, XH-A1 make worse DV. Best result for "sd video" is first shot HDV, than make downconversion (by a sequence of still).

But I can be wrong (hope .-) ).

Bill Busby · May 7th, 2007, 03:17 PM

Pavel, your example & Roland's aren't anywhere near the same contrast/color saturation & can't really be compared... at least that's the way I look at it.

Notice the blocking artifacts where indicated. It's quite similar to Roland's with the blue pants.

Bill

Dick Nelson · May 7th, 2007, 09:57 PM

Make a custom preset with reduced sharpness, HDF and DHV values. This reduces some of the enhancement and will get rid of the stair-stepping on strong verticals and diagonals. Try different amounts of each until you get the amount of smoothness you like. The artifacts you're seeing will be less of a problem with less enhancement processing.

Maksim Yankovskiy · May 8th, 2007, 12:46 PM

Here's what I want to try one day not too long from now.

HD image is 1440x1080 at 1:33 aspect.
DV image is 720x480 at 0.9 aspect.

I want to try shooting in HDV, but to frame my shot such that the subject and the important part of the image I am capturing are occupying only a small portion of the HDV image. The rest of the HDV image will be wasted. I will then transfer the HDV stream to computer, and instead of resizing the image (which, inherently, introduces artifacts), I will crop the image to 720x480 discaring the part of the frame I don't need (I can do that, because I shot this way). So, there will be no resizing, hence, less artifacts.

Of course, this approach creates problems for wide-angle shooting situations.

Evan Donn · May 9th, 2007, 02:21 AM

That looks more like the result of red & blue channel color subsampling in DV (4:1:1) - that's why you only see it in very saturated blue or red areas and not in other parts of the picture. I see this all the time in DV - it's often more apparent when looking at the frames on a computer vs. on playback as the analog outputs on most DV cameras seem to smooth this out on playback.

May 7th, 2007, 05:19 AM	#1
Roland Oeller New Boot Join Date: Dec 2006 Location: Austria Posts: 13	XH A1 compresses in HDV then writes in DV to tape? Hi! I have now shot my first project in SD DV (after shooting some stuff in HDV) and while capturing I noticed that the image looked somewhat MPEG2-compressed. This should not be since I recorded in 4:3 DV 25f. Also I played the tape on an old DV-camcorder so the signal is definitely DV. I captured in Premiere Pro 2 with the usual DV-settings. I suspect that the A1 internally uses the HDV signal as a basis and converts it to DV. The compression is not that obvious but it is there. You may have to download the image to be able to zoom in and see what I mean. http://img143.imageshack.us/img143/1...ession1bm7.png Here is a portion of the image resized to 400%: http://img516.imageshack.us/img516/6...ession2pc9.jpg Look at the lower edge of the pit with high contrast and the edge of the wheelbarrow. There are some obvious compression artifacts there. And look at the blue trousers which show macroblocking. This is not a huge problem for me since I care more about HDV but I find it disturbing that nobody seems to have noticed this yet. A DV-image should not look like this, right? Maybe it is only something the PAL-Version of the camera does? Maybe it´s the Frame-mode? Or are my eyes playing a trick on me? Any thoughts?

May 7th, 2007, 10:22 AM	#3
Pavel Sedlak Major Player Join Date: Feb 2007 Location: Prague Czech Republic Posts: 263	re I have the same experience, DV have some mpeg2 compression residue. When you make SD mpeg2 (from this DV), it is more noticable.

May 7th, 2007, 10:37 AM	#4
Don Palomaki Wrangler Join Date: Sep 2001 Location: Northern VA Posts: 4,488	From a RFC: "The DV format only uses the DCT compression technique within each frame, contrasted with the interframe compression of the MPEG video standards. All video data, including audio and other system data, are managed within the picture frame unit of video." "The DV video encoding is composed of a three-level hierarchical structure. A picture frame is divided into rectangle- or clipped-rectangle-shaped DCT super blocks. DCT super blocks are divided into 27 rectangle- or square-shaped DCT macro blocks." So with the DV format, upon close examination, you can expect to see some macro block artifacts. However, because each frame stands alone compression wise, they may not be as apparent in video (as opposed to individual frame grabs). __________________ dpalomaki@dspalomaki.com

May 7th, 2007, 12:47 PM	#8
Pete Bauer Wrangler Join Date: May 2003 Location: Eagle River, AK Posts: 4,100	The sawtooth along high contrast edges in Pavel's picture look more like interlace artifact from a slowly panning camera than compression artifact. __________________ Pete Bauer The most beautiful thing we can experience is the mysterious. It is the source of all true art and science. Albert Einstein Trying to solve a DV mystery? You may find the answer behind the SEARCH function ... or be able to join a discussion already in progress!

May 7th, 2007, 03:17 PM	#10
Bill Busby Major Player Join Date: Dec 2006 Location: Los Angeles, CA Posts: 959	Pavel, your example & Roland's aren't anywhere near the same contrast/color saturation & can't really be compared... at least that's the way I look at it. Notice the blocking artifacts where indicated. It's quite similar to Roland's with the blue pants. Bill Attached Thumbnails

May 7th, 2007, 06:02 AM	#2
Mikko Lopponen Major Player Join Date: Mar 2006 Location: Helsinki, Finland Posts: 393	Hard to say. DV-compression does use mjpeg and it also causes those jpeg-compression artifacts. Shoot something where there's A LOT of motion and it should be apparent if it's gone through the mpeg2 encoder.

May 7th, 2007, 10:50 AM	#5
Bill Busby Major Player Join Date: Dec 2006 Location: Los Angeles, CA Posts: 959	My take on this. Blow ANY DV signal up to 400% & you'll see the same results. The original pic you posted looks fine here. Bill

May 7th, 2007, 11:57 AM	#6
Pavel Sedlak Major Player Join Date: Feb 2007 Location: Prague Czech Republic Posts: 263	Blow ANY DV signal up to 400% & you'll see the same results. http://www.videoproduce.cz/images/DV_400pct_vx2000.jpg it's not look like the same result (specially about edges)

May 7th, 2007, 12:45 PM	#7
Bill Pryor Contributor Join Date: Jan 2003 Location: Kansas City, MO Posts: 4,449	How does it look when playing back, not from a freeze frame?

May 7th, 2007, 02:51 PM	#9
Pavel Sedlak Major Player Join Date: Feb 2007 Location: Prague Czech Republic Posts: 263	Pete, you are right. Only interlacing, nothing else (like XH-A1 dv). I mean that difference is too big, XH-A1 make worse DV. Best result for "sd video" is first shot HDV, than make downconversion (by a sequence of still). But I can be wrong (hope .-) ).

May 7th, 2007, 09:57 PM	#11
Dick Nelson Regular Crew Join Date: Aug 2006 Location: Silverthorne, CO Posts: 36	Make a custom preset with reduced sharpness, HDF and DHV values. This reduces some of the enhancement and will get rid of the stair-stepping on strong verticals and diagonals. Try different amounts of each until you get the amount of smoothness you like. The artifacts you're seeing will be less of a problem with less enhancement processing.

May 8th, 2007, 12:46 PM	#12
Maksim Yankovskiy Regular Crew Join Date: Apr 2007 Location: San Francisco, CA Posts: 114	Compositing Here's what I want to try one day not too long from now. HD image is 1440x1080 at 1:33 aspect. DV image is 720x480 at 0.9 aspect. I want to try shooting in HDV, but to frame my shot such that the subject and the important part of the image I am capturing are occupying only a small portion of the HDV image. The rest of the HDV image will be wasted. I will then transfer the HDV stream to computer, and instead of resizing the image (which, inherently, introduces artifacts), I will crop the image to 720x480 discaring the part of the frame I don't need (I can do that, because I shot this way). So, there will be no resizing, hence, less artifacts. Of course, this approach creates problems for wide-angle shooting situations.

May 9th, 2007, 02:21 AM	#13
Evan Donn Major Player Join Date: Feb 2006 Location: Philadelphia Posts: 795	Color subsampling? That looks more like the result of red & blue channel color subsampling in DV (4:1:1) - that's why you only see it in very saturated blue or red areas and not in other parts of the picture. I see this all the time in DV - it's often more apparent when looking at the frames on a computer vs. on playback as the analog outputs on most DV cameras seem to smooth this out on playback.