GH1 firmware research volunteers required

Sorry, I didn't realize it wasn't the right thread, as this seemed to be the thread that deals with sensor refresh rate hacks. Could you please me direct me to the right thread then? May be worth listing it/them also in the first post (like bitrate discussion... go to http:/...., refreshRate discussion... go to http:/...., etc.)

Thanks.
 
I guess I'm 1 of 3...I was here to volunteer.

I guess I'm 1 of 3...I was here to volunteer.

I'm trying to figure out what to get..note that was my first post. I'll figure it out, hoping someone else started work on G2?


Guys, all three last messages posted in wrong thread.
Will be deleted soon.
Please, post in correct threads.
 
Anyone able to get 24p native playback in conjunction with tweaking the Audio encoding bps? I have tried both PAL/NTSC as well as the range within Audio Encoding Bps- anything but the original default will not allow playback in camera.

Can anyone confirm this?

Nope at this stage this doesn't work yet. 192kbs is plenty especially if you are using a rode video mic which is mono.

But would be nice if it did work!
 
Awesome work tester.

It would be very useful for 3D shooting if you could swap the recording dimension around, eg. record in 1080x1920. This would allow you to use two cameras rotated 90 degrees and so get the two lenses closer together. If you use the normal orientation, the closest you can get the GH1 lenses (with two bodies next to each other) is around 11cm centre-to-centre, and this is much wider than the average human eye separation (aka as 'interocular') of ~6.5cm. Using the cameras sideways gets the lenses closer to that. (I've prepared some graphics to show various GH1 3D configurations, if anyone's interested PM me).
 
GF1 test report:

i tried again and again and again with all kinds of settings, but the 1920*1080 mjpeg never work. just recorded 1-2 seconds and quite every time even the bitrate just only 43Mbps. i'v changed the mjpeg quality settings as 156 128 110 98 and all tables 24, no 4:2:2 patch, but still the same.

now i changed mjpeg size to 1440*800 with the 4:2:2 patch, quality settings 384 320 256 200, and it works fine. the bitrate is about 64Mbps.
 
Is there a good post that explains what each of the mjpeg E1-E4 settings does? Do I need to set them all?


I want to use this question to ask tester13 a question I can't get out of my head:

You say that you think that the other three values E2-E4 are "fallback" values in case of buffer problems. Is it possible that e.g. E1 is the quantisation table for luma, E2 and E3 for the two chroma parts and E4 for the thumbnail? How does the part for still photo JPEG compression look in the code? Are there also four settings? What are Panasonics defaults there? Just a thought and maybe a dumb one.
 
tester13, my question to you is -- is there any way you can optimize the bandwidth to prioritize the I-frame? I had one clip which had really bad "pulsing" -- too much detail for the I-frame, so every half-second it would "glitch" for a frame or two while the rest of the GoP "caught up".

I think that especially with the higher bitrates for the overall GoP that are being achieved, that we could really see some benefits from allocating more bandwidth to the I-frame so that the I-frames are cleaner. Is that possible?
 
I want to use this question to ask tester13 a question I can't get out of my head:

You say that you think that the other three values E2-E4 are "fallback" values in case of buffer problems. Is it possible that e.g. E1 is the quantisation table for luma, E2 and E3 for the two chroma parts and E4 for the thumbnail? How does the part for still photo JPEG compression look in the code? Are there also four settings? What are Panasonics defaults there? Just a thought and maybe a dumb one.

Here is the answers:
1) I don't think that your theory about tables is right.
But it can be easeally checked by tests.
2) One interesting thing is that as you know table have special value before each quality-table pair. And 320x240 have less such pairs. Plus it is only 422 resolution bu default.
3) AVCHD and pictures seem to refer to 1024 table values, but do not have
any similar quality-index tables. But it is my curret limited view.
I'll move it to MJPEG encoder thread after some time, ok?
 
tester13, my question to you is -- is there any way you can optimize the bandwidth to prioritize the I-frame? I had one clip which had really bad "pulsing" -- too much detail for the I-frame, so every half-second it would "glitch" for a frame or two while the rest of the GoP "caught up".

I think that especially with the higher bitrates for the overall GoP that are being achieved, that we could really see some benefits from allocating more bandwidth to the I-frame so that the I-frames are cleaner. Is that possible?

Today I can't do it.
But in future it seems possible.
 
tester13, my question to you is -- is there any way you can optimize the bandwidth to prioritize the I-frame? I had one clip which had really bad "pulsing" -- too much detail for the I-frame, so every half-second it would "glitch" for a frame or two while the rest of the GoP "caught up".

I think that especially with the higher bitrates for the overall GoP that are being achieved, that we could really see some benefits from allocating more bandwidth to the I-frame so that the I-frames are cleaner. Is that possible?

I noticed the same thing on very highly detailed scenes, with increased bitrates for AVCHD the half second pulsing seems much more noticable. Obviously the increased bitrate for the P - frame makes the codec to catch up every half second, but inbetween you can see how it looses detail and catches up again.

P.S.: The crazy part of the above statement is that these details which are now pulsing were not been seen all with the original GH1! The IQ now is just amazing!
 
Last edited:
GF1 test report:
...
now i changed mjpeg size to 1440*800 with the 4:2:2 patch, quality settings 384 320 256 200, and it works fine. the bitrate is about 64Mbps.

thanks I tried this settings 1440 * 1080 and works great, I used table 24 and no 4:2:2. Using a cheap class 6 and no problem. The only thing is that I can´t play back in camera. (you?) mabey changing the table setting can resume playback but I´m lost with this table settings is it like lottery?

I can see a little bit more detail in resolution comparing to the pimped avchd.

this with 25p is incredible.
:-)
 
Tester13, there is a debate on whether video on the GH1 is set sRGB only. People claim that changing in the menu to AdobeRGB yields better results. Is there a way for you to check and see if indeed sRGB is the only one allowed for video. Thanks.
 
PTool 3.36d released (development version for testing only)

http://www.pentax-hack.info/firmware/pan/ptool3d.zip

Added Version compare patch.
Short description: You need to set version increment as usual first time then using this patch. But since next upgrade you can omit increment at all, and leave this patch only.

This is risky patch, as can make body non-upgradable. Probability of this is very low. Bit it can happen, as it is untested patch. So, one brave tester is required.
 
hi tester13
here is some minor problem with native 25p footage
final cut interpret it as interlaced - upper field first
thanks
 
Tester13, did you say you could 422 to AVC?

Tester13, did you say you could 422 to AVC?

Would be interesting? also be interested to know if you think the adobeRGB is "flatter" profile to SRGB in video mode. It looks completely different to me and records differently.
 
Back
Top