GH1 firmware research volunteers required

[Kholi]

Just incase there was any doubt about it being upscaled 720p, I ran some some chart tests on the 1080p MJPEG mode. All stills are 1080 off camera or 720 smooth bicubic upscaled to match, then all images cropped back down to 720. Some of the tests were a bit redundant, for instance there was no expected change in AVCHD unpatched and patched, but it's there for posterity.

1) AVCHD 1080p Unpatched

2) AVCHD 720p Unpatched

3) MJPEG 720p Unpatched

4) AVCHD 1080p Patched
VidBuff: 40 VidRate: 32000000
OverRate: 35000000 Limit Rate: 40000000

5) MJPEG 1080p Patch (2 Takes)
MJPEG Size 1080 -> 1920
MJPEG Encoder 1080 -> 1920
MJPEG Quality: 720>512>428>400
MJPEG Table: 24>24>24>24

Zipped Stills:
www.jerseyforged.com/sampleclips/MJPEG1080/1080p_Compared.zip

When switching back and fourth you can clearly see the increase in sharpness/detail in the native/AVCHD 1080p shots... although the MJPEG pre-encode-upscaled 1080p is still a small magnitude sharper then the native 720p.

AVCHD 1080p vs MJPEG 1080p:



Mind you, the increased sharpness in MJPEG 1080 is a tradeoff with squeezing more resolution into a given bitrate. In this example, the abundance of test frame is low detail (as was most of what I croppped out) and the bitrate was set to "100mbps then stall" settings. So you expect to get sharper results, but with more jpeg compression in more detailed scenes. Even this shot clearly has more compression artifacts vs the AVCHD.

Here's what I'm noticing about that chart, and maybe we just need to see another resolution chart with the same comparison: the AVC-HD exhibits more aliasing than the Photo JPEG material.

Is that just me?

[ChipG]

Time to get those fine striped shirts out that everyone burned.

[GrgurMG]

Here's what I'm noticing about that chart, and maybe we just need to see another resolution chart with the same comparison: the AVC-HD exhibits more aliasing than the Photo JPEG material.

Is that just me?

Hrm, well.. for starters there could be differences is how the sensor is aligning with the chart in each shot, which could cause abnormalities. Even though I was being careful with the tripod... pretty much stepping on the ground near it would cause a change in the fine line distortions. Also, obviously the MJPEG is conciderably softer since it's upressed in camera, so the AVCHD has more fine detail to exhibit aliasing on.

[philip bloom]

wonderful Job Tester 13...my gh1 is now out of the dusty cupboard and back in my hands!

[pianoboy]

I noticed that too... in the circles at the top corners, the AVCHD lines seem to go curvy towards the center, while the MJPEG lines stay straight.

[alignment1]

has anyone had any success w/ a class 4 card?? (16GB Sandisk Ultra II Class4 15mb/s)

I just want a slightly higher bitrate and native 24p 720/1080--I guess it depends on scene detail eh??

Z

[bumkicho]

I only have class 4 cards. I have no problem using higher bit avchd.

[alignment1]

I only have class 4 cards. I have no problem using higher bit avchd.

GREAT--
next question- I downloaded Winebottler to run windows on osx. when I download Ptool- do I have to do anything special to get it to open- or do I just click on it or something??

I tried running 'Crystal Disk Mark' to test disk speed but kept getting errors trying to install/open it.


Any Help Please

[Kholi]

Hrm, well.. for starters there could be differences is how the sensor is aligning with the chart in each shot, which could cause abnormalities. Even though I was being careful with the tripod... pretty much stepping on the ground near it would cause a change in the fine line distortions. Also, obviously the MJPEG is conciderably softer since it's upressed in camera, so the AVCHD has more fine detail to exhibit aliasing on.

Ah that would be a flub if so. Thanks for showing, either way. I love how the AVCHD 1080/24 stuff looks, can't wait for Photo JPEG 1080/24 to hit, and then the video out.

Looking for my GH-1 right now.

[Barry_Green]

Okay, I just loaded up PTool using the "D" settings, but I modified the buffer to be 60, so it's as big as the maximum bitrate. Then I took it to the forest/lake park with some ducks. Shot a whole bunch of 1080/24p native mode. It's absolutely awesome. This patch is the real deal, and I totally love that there's no pulldown! I shot scenes that the GH1 has choked on before, including high-shutter speed fast pans past a forest of pine trees, and the new "D" settings handled it just fine. It's much sharper than it ever was before.

I also tried the MJPG patch, but I can only get a couple of seconds of recording before there was a card write speed error message. Using a Lexar Platinum card, which is only rated at Class 6 but CrystalDiskMark rates it at 9.7 megabytes/sec. MJPG wasn't really useful at this point. I'll have to get a Lexar Professional card to do more testing.

The AVCHD mode did run into a write-speed problem once, but only once out of 30 clips.

I also shot a chart; I'll report back on that after evaluating the footage.

[Kholi]

Okay, I just loaded up PTool using the "D" settings, but I modified the buffer to be 60, so it's as big as the maximum bitrate. Then I took it to the forest/lake park with some ducks. Shot a whole bunch of 1080/24p native mode. It's absolutely awesome. This patch is the real deal, and I totally love that there's no pulldown! I shot scenes that the GH1 has choked on before, including high-shutter speed fast pans past a forest of pine trees, and the new "D" settings handled it just fine. It's much sharper than it ever was before.

ABout time! You took too long! Please up some stuff. Thanks for going out into the field and reporting.

I also tried the MJPG patch, but I can only get a couple of seconds of recording before there was a card write speed error message. Using a Lexar Platinum card, which is only rated at Class 6 but CrystalDiskMark rates it at 9.7 megabytes/sec. MJPG wasn't really useful at this point. I'll have to get a Lexar Professional card to do more testing.

Okay, so basically when I get a GH-1 again, I need to grip two of the highest class cards out there. Good to know.

The AVCHD mode did run into a write-speed problem once, but only once out of 30 clips.

I also shot a chart; I'll report back on that after evaluating the footage.

Thanks Barry_G. Can you do an MJPEG vs AVCHD test for moire/aliasing? Something quick and obvious?

[Barry_Green]

Here's what I'm noticing about that chart, and maybe we just need to see another resolution chart with the same comparison: the AVC-HD exhibits more aliasing than the Photo JPEG material.

Is that just me?

Yes, it's just you -- because shooting the same chart, there's not a doubt in my mind that the AVCHD is a far more "correct" representation of the image. The MJPG looks to be some sort of uprez.

I'll show you just a bit of what I'm talking about, the central section of the chart with the concentric circles. All you should see are concentric circles, but you'll see that there's a weird scaling pattern in the 1080 MJPG.


For my money, it's AVCHD for the win. It's already 24p, and the 1080 is better "real" 1080 rather than a scaled 720p MJPG image that's running at 30p.

tester13, I apologize for suggesting that you should forget AVCHD and focus on MJPG. You didn't give up, and you came out with a breakthrough patch that is totally fantastic. I am now officially switching sides, and saying AVCHD for the win. It looks incredible, and it's already at 24fps, and it's native 24p. Well done!

[bumkicho]

wonderful Job Tester 13...my gh1 is now out of the dusty cupboard and back in my hands!

Good to see it found its way back into your hands. Hope to see a video or two from you with this new patch.

[e-steve]

The excitement is clearly building with Barry and Philip getting into the action!!

Awesome work guys.

[Barry_Green]

ABout time! You took too long! Please up some stuff. Thanks for going out into the field and reporting.

I can't tell you just how good these ducks look. I remember being so flummoxed by the seagulls I videod, some of them looked pretty good and some were mushy, and I had to wrestle with the sharpness and contrast to try to get any detail on them. Well, let's just say that it ain't that way with the ducks!

Okay, so basically when I get a GH-1 again, I need to grip two of the highest class cards out there. Good to know.

I'm already searching the online tests. It's leading me towards the Lexar Professional, until the new Fuji cards drop in November.

Thanks Barry_G. Can you do an MJPEG vs AVCHD test for moire/aliasing? Something quick and obvious?

Can't do a fully proper one until I get a faster card that'll support more than a second or two. But right now, as of what I've already shot, I don't really see much a point in continuing with the MJPG.