It was suggested earlier that some tests on antenna, cable, VHS, or some other non-ideal source would be of value. I agree. "Popular Photography" has always done camera lens tests in a very controlled environment for contrast, resolution, flare, pincushion, etc. but photographers do not shoot test patterns, so they follow up with some actual photos. This is an attempt to do the same.
The problem with cable, satellite, and antenna is that the source is fleeting, making it difficult to capture the exact sequence in various modes. For this test, a VHS tape version of "The Matrix" was chosen.
 Forgot to mention the VHS version of "The Matrix" was played back on a Panasonic PV-V4524S VHS player to one of the composite inputs ( IN1 ) of the ES10. "The Matrix" was a commercial tape but apparently not copy protected.
The tests include VOB files in XP, SP, LP, and FR3 ( flexible record - 3 hour ) modes. The sequences include a fast action one and a medium one. The video clips are about two seconds long.
Fast action sequence, XP mode, VHS source.
Fast action sequence, SP mode, VHS source.
Fast action sequence, LP mode, VHS source.
Fast action sequence, FR3 mode, VHS source.
Medium action sequence, XP mode, VHS source.
Medium action sequence, SP mode, VHS source.
Medium action sequence, LP mode, VHS source.
Medium action sequence, FR3 mode, VHS source.
Gshelley61 picked an image from the fast action, LP mode sequence with the DVD source. This test continues that except that the frames have not been resized to 720 by 480 but left in the original 704 by 480 format. The source in this case is VHS of course. I was going to upload in png format but the size of the files in 704 by 480 format was about 300 KB. Too large to upload so they were changed to jpg with an 85% quality setting.
Fast action, XP mode, VHS source.
Fast action, SP mode, VHS source.
Fast action, LP mode, VHS source.
Fast action, LP mode, DVD source from gshelley61.
Fast action, FR3 mode, VHS source.
These images are from the medium action sequence.
Medium action, XP mode, VHS source.
Medium action, SP mode, VHS source.
Medium action, LP mode, VHS source.
Medium action, FR3 mode, VHS source.
I was hoping to make this testing easier with mpg2cut2 freeware but in my opinion, it does not work as well as mpeg2schnitt. Mpg2cut2 works on VOB's at the gop level so Vobrater, Mpeg2schnitt, and Ifoedit would not have been needed. One problem with mpg2cut2 is that it does not accurately describe the size of the file that will be created. Seems to go in one MB steps. That is not so easy to use with the 2 MB file size limit for this site.
As before, those who do not want to burn a dvd with the VOB files can play them on Windows Media Player by downloading and changing the extension tfrom .vob to .mpg.
+ Reply to Thread
Results 91 to 103 of 103
The issue with Panasonic is not the mere fact of blocks, but the movement of said blocks from frame to frame, which creates a patchwork pattern that is ONLY noticed because of the changes of block color.
Though other machines can make blocks from time to time, they are usually static in color, and therefore less noticeable, because they do not draw attention to themselves while briefly on screen. Most machines simply add some mosquito noise when the encoder is overloaded and bitrate is inadequate. Panasonic goes to blocks.
While these still and super-short clips make for an interesting read, it's missing this bigger picture. Such things can only be viewed in a 30-second to 1-minute duration, not a couple of frames or handful of stills.
Not to demean the tests above, but I just want this to be pointed out and remembered. The above tests easily show how the Panasonic is adding trailing mosquito noise in SP mode, and blocks beyond that. When put into a fluid motion of clips, it only gets worse.
The site limitations make 30 second video clips out of the question. I think we are near the limits of what can reasonably be demonstrated with any recorder within those limits. It has been quite a learning experience. Sorry I cannot show the VHS source quality. The tape had to be returned.
Thanks for all of your efforts trhouse. From those clips I think it does a solid job at 3 hours and less, but at 4 hours it isn't doing too good. They should have switched to 352x480 at about 3 hours or so as 4 hours is alot of content to fit at full resolution, no matter HOW GOOD the encoder is.
Also, how close do the Panasonics come to filling out the DVD? I thought I'd read that they only fill out to about 4gb, while my Pioneer 420 fills the discs to 4.36gb/~4460mb. While 4gb vs 4.36gb doesn't sound like that big a difference, it's about 10%. That's about 20 extra minutes at LP.
Regarding the LP mode performance. Until gshelley61 posted the still image taken from the fast action LP mode sequence, I had not noticed the macroblocks in the sequence itself viewed as video. It is a good choice to use for comparing how the different modes affect the macroblocks. To put it more in perspective, here are five frames from that sequence including the image mentioned. This entire sequence is about 83 milliseconds long.
This the gshelley61 frame
I checked the size of three disks recently burned in FR mode. Windows reports them to be 4 GB or about 4,300,000,000 bytes.
These tests have gone a lot further than originally intended but it has been a pretty valuable education. It would have been better to title this thread, "Tests you can do with freeware on your recorder". It seems only incidental that the ES10 happened to be the test vehicle.
I do appreciate your note of thanks.
OK - I got hold of an ES10 and have done a few tests, as well. I used the s-video output of my Pioneer DVR-210 DVD player as the source and captured to -RW in Video format. After doing some THX test pattern captures, I recorded from a 4:3 DVD of the Paul Newman classic movie "Cool Hand Luke". It has some moderate action scenes and film grain (which can be a challenge for encoders).
The ES10 does a great job in 1 hour XP mode, but noticeable macroblocks and other compression artifacts begin to appear in the 2 hour SP mode. The 4 hour LP mode, which maintains 704x480 capture resolution, has a hard time with motion and produces lots of macroblocks, color banding, etc. However, when scenes settle down or are still, the LP mode looks pretty good due to the higher resolution. Test patterns looked fine, for example.
I recorded the same tests with my JVC DR-M10 to compare. The Pioneer DVR-210 was also the source DVD player.
Frames were grabbed with VirtualDubMod and resized to 640x480 (4:3) and saved to JPG with IrfanView. Panasonic captures at 704x480 by cropping 8 pixels on each side. The JVC captures at 720x480.
This first frame is an action scene where the Captain strikes Luke with a blackjack...
Panasonic DMR-ES10, XP mode
Panasonic DMR-ES10, SP mode
Panasonic DMR-ES10, LP mode
Here's the JVC recordings of the same scene:
JVC DR-M10, XP mode
JVC DR-M10, SP mode
Although there are some compression artifacts in the JVC's SP capture, it looks better than the Panasonic in the same 2 hour mode.
JVC DR-M10, LP mode
Naturally, the 352x480 capture resolution that JVC uses in their LP mode makes the picture softer and less defined, but there a fewer macroblocks than the Panasonic LP recording. I wouldn't use LP mode on either machine, personally.
Here's a still scene from the same movie:
Panasonic DMR-ES10, SP mode
JVC DR-M10, SP mode
Again, the JVC has a little bit smoother picture with fewer artifacts. Both machines are employing video noise reduction to deal with the film grain... this is where the JVC has an edge over other brands. Their noise reduction circuit does an outstanding job.
Those are some good comparisons. I had been thinking earlier that the ES10's LP mode tests did prove a bit useful. The full resolution seems to make animation a good candidate for LP capture; something that I had not thought of before. Animation has less detail and large patches of color often unchanging. "The Matrix" movies would be a poor candidate for LP mode.
The ES10 macroblocks are more visible in the SP mode images. There is greater color gradation between the blocks. I was expecting the M10 to have fewer macroblocks, but the density and placement of the blocks seems pretty similar although the smaller color changes between them make the boundaries less distinct. These blocks from either recorder may not be easy to see in a video clip. It took some seconds for me to see them in the stills and this sequence is probably a fraction of a second in length.
The last two images are so close that they would be hard to identify with my LCD monitor. Images act as if the contrast is changing as they are scrolled from top to bottom. The closeness of color, hue, saturation, and sharpness of all the images is very good.
Can you post an LP image from the M10 for comparison to the LP from the ES10?
Done... I added some more frame grabs to my post above.
The M10 LP mode macroblocks are fewer but look twice the width of the ES10 ones. There must be a relation between the horizontal resolution and block size. The M10 color gradation between macroblocks remains less making them less distinct.
I must not know what to look for in the last three images. The original, ES10, and M10 look identical.
I suspect that these tests are in some way a proof of Nyquist's Theorem.
"Nyquist's theorem: A theorem, developed by H. Nyquist, which states that an analog signal waveform may be uniquely reconstructed, without error, from samples taken at equal time intervals. The sampling rate must be equal to, or greater than, twice the highest frequency component in the analog signal."
Originally Posted by trhouse
My monitor is a Samsung as well ( 171V ). I was looking in the wrong place in the last three pictures. The differences are in the sky. The artifacts in the sky I mistook for heat shimmer, but both recorded images do show a smoother sky than the original.
That is a good explanation regarding the size of the macroblocks vs horizontal resolution. Before you posted these images, I thought macroblocks were flaws that just occurred, but now I suspect it is a tool used by the creators of one pass VBR encoders to handle fast action scenes by creating squares with fixed content to decrease the amount of information to be processed in a given amount of time. Much as animation with its very large patches of fixed color requires less of the encoder. The size of the squares created by both recorders in terms of pixels is pretty similar although the designers of the JVC encoder handle the color transistions between blocks better.
Yes, in terms of color, detail, sharpness these machines have much in common. In general, I am more impressed by the number characteristics which both machines share almost identically than the few in which they differ.
Actually, most DVD recorders do a very good job at the higher bitrate settings. There are some quality differences in how analog video is processed and sampled, and some encoding chips do a better job than others. As others have mentioned, picking out the right DVD recorder really involves digging deeper into the various features and operational characteristics of the models you are interested in, then choosing one that best suits your needs.
I think you are right in your above comment.
But, I think that the main reason why all the variances between
the users experience/resutls, are because the (shall I say) "Holy
Grail" of player devices (vcrs; dvd players; etc) are not being
used in such users' demonstrations.
So far, I have narrowed down the players. And, it looks to me,
like the Sharp has it. From another topic, a Sharp 600 dvd player
was used. And, in another thread, a Sharp nameless model no was
used also. And, it seems that Sharp, has the color space/IRE
worked out closest to the source.
- I say, lets beat around the bush, and we all go out and get ourselves
a player put out by Sharp, and re-do *ALL* our tests
- What do you say ??