MJPEG Encoder Research, no questions here

[Vitaliy Kiselev]

Cut after E1 doesn't seem to have an effect.

Try two other cut patches.
Main quality definition is E1, so testing must include really big changes in detail.

[Martin Koch]

CUT AFTER E2
Q: 128, 110, 100, 92
T: 133, 146, 150, 155
Low detailed scene: 83.09 / 0.6 mbps / PLAYBACK OK / MOVIE OK
High detailed scene: 19.16 / 37.34 mbps / PLAYBACK OK / MOVIE OK
4:2:0
Clip recording time display: 8m 20s

CUT AFTER E3
Q: 128, 110, 100, 92
T: 133, 146, 150, 155
Low detailed scene: 83.09 / 0.6 mbps / PLAYBACK OK / MOVIE OK
High detailed scene: 19.16 / 36.80 mbps / PLAYBACK OK / MOVIE OK
4:2:0
Clip recording time display: 8m 20s




CUT AFTER E1 Q400 / T4
Q: 400, (110, 100, 92)
T: 4, (146, 150, 155)
Low detailed scene: 100 / 64.08 mbps / PLAYBACK NO / MOVIE OK
High detailed scene: 96.39 / 180.28 mbps / PLAYBACK NO / MOVIE CANCELLED
4:2:0
Clip recording time display: 2m 39s

CUT AFTER E1 Q300 / T4
Q: 300, (110, 100, 92)
T: 4, (146, 150, 155)
Low detailed scene: 100 / 62.83 mbps / PLAYBACK NO / MOVIE OK
High detailed scene: 83.44 / 137.28 mbps / PLAYBACK NO / MOVIE CANCELLED
4:2:0
Clip recording time display: 3m33s

CUT AFTER E1 Q200 / T4
Q: 200, (110, 100, 92)
T: 4, (146, 150, 155)
Low detailed scene: 100 / 59.76 mbps / PLAYBACK NO / MOVIE OK
High detailed scene: 83.44 / 96.15 mbps / PLAYBACK NO / MOVIE OK
4:2:0
Clip recording time display: 5m 21s

[Lpowell]

Try two other cut patches.
Main quality definition is E1, so testing must include really big changes in detail.

Thanks for the new patches, Vitaliy. Can you give us a short technical description of what effect the new Cut after E1-3 have on the MJPEG encoder?

[garhodes]

A thought: I wonder if removing audio record from the firmware would allow for significant gains in image bitrate... I have noticed that increasing audio sampling will cause the camera to freeze up, perhaps the inverse is true-- dumping audio bitrate or removing audio record would free up bandwidth and processor in mjpeg...

[park25]

When I first opened the .mov with JPEGsnoop it said, "NOTE: File did not start with JPEG marker. Consider using [Tools->Img Search Fwd] to locate embedded JPEG."

So I searched forward 2 or 3 times until it found a frame.

I think JPEGSnoop has found a thumbnail image recorded simultaneously by the GH1, not a frame inside the Mov file.

Deleting the jpg thumbnail file will support my doubt.

[spamrakuen]

Hi guys, I don't know if it will be useful (for you) or not, but I've made a small application for extracting JPGs from inside MOV files.

Application can be found here;
http://www.dvxuser.com/V6/showthread.php?t=217884

It uses direct stream copy, so no recompresion is done and it's very fast.

[Eftegarie]

Finding the Relationship between the MJPEG NUMBERS LOGIC on GH1 Hack project

More tests and still no clue.

Dear Martin and other Pros,
Because many Pro's (including myself) don't have a clear clue to the EXACT working logic between the Quality and Table NUMBERS, I have started special interest in these eight numbers and hope to find the truth behind the relationship.

For this I have begged our "data expert" here in the office to take an alternative look into these numbers. He suggested to VISUALISE them as well as COMPARE them with at least 10 other good settings. From this he said, the logic will slowly appear... After comparing five favourite famous settings we stopped, so that you can reply what complementary good settings are missing.

These graphs SHOULD be seen SIDE BY SIDE, to compare them DIRECTLY:
1 * Martin Kochs Settings (famous for impeccable good tests, far edges)
2 * Pappas Settings (famous)
3 * Andrew Reid settings (famous artistis video, makes us forget the numbers)
4 * ModulR (pushing things to the far edge)
5 & Default Panasonic (most reliable yet)
6 -10 * ... need 5 more complementary good settings before continuation...

RESULTS SOFAR:
http://aster.nu/gh1.php

These graphs tend to show the RELATION or DISTANCE between Qx/Tx in relation to each other as well as the other three input terms. Bit complex to explain to newbies I guess. NEW graphs will be added in the current row of five, all in one clean clear and easy to find place, as soon as you have your reliable favourite settings, post that valuable data in here, and I will beg my colleage again to put it into visually representable graphs. After that we ill individually connect the ten settings graphs, with the uality and reliability outputs each of the settings.

I will keep coming back to this place reading your extreme quality data that still is reliable enough on class 10 cards.

Power to the people
Cheers, Eftegarie

[crunchy]

Have you mixed the numbers? Check graphs. I suppose that green and violet numbers should be the opposite as they are now. Since some numbers are very small (4) and other large it would be easier to draw them in logarithmic scale?!?

Edit: Are Martin Koch's and Panasonic's settings the same (according to graphs?)?

For example see attached log-scale graph.

[Eftegarie]

...Since some numbers are very small (4) and other large....

Thats what we also thought, at first. But the experts have far far other ways of reasoning...
From what i understand the images should be right. I havent made them, some wise math nerds in the office figured it out during and some time after lunch. I put the question on their table, as a proud GH1 owner, and they asked for the stuff above. Our data analyst intern said that: If distance between the items is non-linear, then the amount is best visualised as a cross-relationship, in order to emphasize the Qx/Tx distances from each other. Also, the size of the balloons show their numeric content (so don't look at the heights of the balloons) 4 is small balloon, 250 is big ballon. From what i understand the RELATION is what we are after, NOT the exact numeric value in a linear or logarithmic or any other scale for that matter. THATS why the logic is so mysterious... we have to find out... Since tome use T=4 and some T=100, BOTH producing SAME high Mbit/s, that amkes your graph not valuable since the relation or distance between each Qx/Tx setting is not apparent, while it is THIS relation which is accountable for the quality differences, which in the end what we're after: HiDef lossless video...

Keep sending valuable data please Martin Koch and other testers, and don't waste your time, my time or our data analysts who are hungry for more data input by those who went to various extremes producing same high bitrate and reliability. it is THAT data that our data analyst wants.
Cheers!

[Eftegarie]

PROS: Be sure to check AVCHD as according to our quality tests (resolution+contrast in controlled environment) and motion tests (city life in uncontrolled and chancing weather environment with more or less constant speed of cars passing by) it seems that AVCHD is sharper than MJPEG though less smooth in motion!

We achieved AVCHD 61 mbit as averaged(!) output: visibly sharper than MJPEG.
Updating now our settings chart in my office (whenever boss is not around):
http://www.aster.nu/gh1.php

Cheers and happy filming.
Power to the People.

[crunchy]

We achieved AVCHD 61 mbit as averaged(!) output: visibly sharper than MJPEG.

Very interesting. Will have to try with shorter GOPs. Have you ever tried shooting some of the charts (PappasArts or Crunchy's) or with 25FPS? I suppose that you tested with 24pn (native) option?

Added: What does it mean that video "hicks? You said "the video hiks (skips some frames once a second orso) as if the choke limitation of 70mbit doesnt allow to dump all frames...". Have you checked video frame-by-frame and does it still "hicks"? Have you checked it with Stream Parser? Please, can you upload somewhere one of the "hicks" videos?

[modulr]

Interesting that extremely low GOPs are possibly working now. Before a setting of 1 or 2 was pretty much guaranteed an instant crash (was this native mode?).

[sasi_yaad]

We achieved AVCHD 61 mbit as averaged(!)

Are you sure there is no mistake here ?
I use similar settings (only on a PAL camera) which result in picks at 55Mbps ....
Can you send a screen shot from Gh13streamparser for that kind of file ?
If you switch to frames mode you can see the average bit rate.

Sassi

[Lpowell]

Interesting that extremely low GOPs are possibly working now. Before a setting of 1 or 2 was pretty much guaranteed an instant crash (was this native mode?).

Even when the camera did not crash, I've spotted frame errors in AVCHD files produced with GOPs of 4 frames or less. These appear to be due to internal codec glitches rather than SD card write speed limitations.

[larryfine451]

I have test shots going from GOP 14 to 7 to 5 to 2 .
In the parser program you can see I frames increasing , and file sizes growing....Big change occurs at GOP5 as pic starts to darken and lose dynamics then at 2 darkens again stream is then odd looking and flattened

All exact same scene no movement in frame file sizes go ~ 21 meg then,75 meg then 129 meg then 158 meg all for the same 40 sec capture I think at GOP 5 we are seeing codec breakdown as this level of low compression is not designed into this hardware platform..
Yes it will do it but not efficiently.. and the picture suffers for it . So far my research seems to want not to push GOP lower than 7 at this point ...GOP 5 seems to make the Pic crush blacks and lose midtones ...more experiments needed .

also bought and tested panasonic gold and sandisk extreme 8gig units

based on cartridge burst mode tester supplied on this forem my results were:

panasonic gold :

56.3 Mbs at 32 k burst
76.6 64k
87.2 128k
83.2 256K

sand* extreme III

80.3 Mbs 32k
102 64k
112.4 128k
112.0 256k

so the sandisk is a lot faster than the panasonic

this is reflected in My Mjpeg resting at 100 to 176 Mbs the panasonic just fails and the sandisk keeps goin on I must say that the 100 Mbs files are also really pushing the hardware as well as blacks seem to be crushed and there is a general sense of stressing the unit and noise generation .. 50 to 75 mbs seems to be about right ..

I am now using Zeis contax G 45 for testing and it will be the definitive unit for clarity as for 35 mm it is considered one of the sharpest lenses wide open ever produced ...so it is overkill for our 1080 p tests...

cheers
Larry