100Mbps Max Latitude Native 24/25p Patch

It's variable bitrate, film some trees with leaves and see what your results are. Use Stream Parser or Media Info.

thanks. but does that mean if i don't shoot a high detailed scene i won't get the best quality?

is the bitrate constant on MJPEG format?
 
thanks. but does that mean if i don't shoot a high detailed scene i won't get the best quality?

is the bitrate constant on MJPEG format?

No, you would still get comparable quality. The extra data is not needed in less-detailed scenes.

I can't say for sure, but the MJPEG probably works the same way as AVCHD with the patch. You can download Stream Parser and it will show the bit rate changes over time:
StreamParser Install.zip
 
No, you would still get comparable quality. The extra data is not needed in less-detailed scenes.

I can't say for sure, but the MJPEG probably works the same way as AVCHD with the patch. You can download Stream Parser and it will show the bit rate changes over time:
StreamParser Install.zip

'm a little bit disappointed. today i finally tested the camera with some high detailed scenes. the highest bitrate i got was 38mbps with this hack, i shot everything in FHD AVCHD 25p. at this scene the camera always froze if i recorded a longer clip then 10 seconds. i have a samsung C10 card.

http://i47.tinypic.com/25f6el3.png (i didn't focus where i wanted but i couldn't see s*%t because of the snow it was soo bright)

that counts as a high detailed scene, right? media inspector says the video steam bitrate is 17.4mpbs, and the scene is very low detailed. is that normal or i screwed something up? i shot other scenes there, but they i didn't like the quality.
 
Sorry to 'grave dig' but I have just bought a mint GH1 and loaded this patch.
My mind is officially blown.

Just wanted to say I am really grateful to the developers concerned. Thank you BIG TIME!
 
'm a little bit disappointed. today i finally tested the camera with some high detailed scenes. the highest bitrate i got was 38mbps with this hack, i shot everything in FHD AVCHD 25p. at this scene the camera always froze if i recorded a longer clip then 10 seconds. i have a samsung C10 card.

http://i47.tinypic.com/25f6el3.png (i didn't focus where i wanted but i couldn't see s*%t because of the snow it was soo bright)

that counts as a high detailed scene, right? media inspector says the video steam bitrate is 17.4mpbs, and the scene is very low detailed. is that normal or i screwed something up? i shot other scenes there, but they i didn't like the quality.

Just to chime in very late, but IMO that is not a particularly detailed scene as far as the encoder goes. Why?

1. No random movement. The codec records changes from one frame to another. That's why foliage and water are high detail, because everything is always changing in the picture. Even in a pan, which would require the codec to work a little bit, the changes from frame to frame are very precise.
2. No color. That is almost a black and white scene. No movement and no color = low bitrate, because it's all you need.

The question is, when you look closely at the image do you see any loss of detail? My guess is no. I still use my GH1 and I never see anything that I'm not happy with the amount of detail in the image. Dynamic range is another story, but detail is always there.

As for the card freezing, my guess is that the card, while class 10, isn't fast class 10. Class 10 just puts it in a category that means it writes greater than 10 MB/s (I believe that's the number). There are various speed checking programs for memory cards. Use one of those to check it, and buy a faster card. You can't go wrong with a Sandisk Extreme Pro 95 MB/s card.
 
Have any idea here? I seem to have a problem, it may have only applied to the motion Jpeg part, because it records at 50 mbs on that setting, however on the AVCHD setting, i could only get it as high as 18 MB, and I tried to get the codac to crash on the AVCHD setting and couldn't, any ideas?
 
* AVCHD 4GB file-spanning for long video takes may not work reliably at high bitrates. For reliable recording of takes longer than about 12 minutes, I recommend selecting the "H" video mode instead of "SH". This will produce average bitrates of about 24Mbps in 720p25/30 modes.

* For extended recording times at moderate bitrates, selecting the "L" video mode instead of "SH" will produce bitrates below 17Mbps.


Just in case anyone is curious this late in the game, using the 100mbps patch on Sunday I did a straight 1 hour and 14 minute recording in FHD of a stage setting (shown below), with a few pans back and forth as the performer moved about the stage but generally stayed like what you see. The screenshot is after I got the spanned files transcoded to DNxHD since Davinci won't support MTS at 1080 resolution (bunch of media offline type of errors and corruption), nor does it support the native AC3 within the wrapper so has to be transcoded anyways (at 720p I can easily pull the video in without transcoding, but won't use the audio, which is fine if I'm dropping in audio from an external recorder anyways).

Only had one error that evening when a projector screen came rolling down and the backstage went completely dark for about 10 or so seconds which caused a write error to show up on screen (something I haven't had a problem with with the 75mbps patch). I'd have to do more runs with this patch but for whatever reason I keep preferring the 'look' I get out of the 75mbps patch, just not sure if I'm losing anything if I transcode the 29.97 interlaced MTS to 24 DNxHD (thread says it's 3:2 pulldown with 24 sensor output) since I'm having to transcode anyways. Course we have a great many new cameras out in 2017, but I'm going to milk this GH1 I got for under 100 for all I can.

brandon-media.jpg
 
PS: Any thoughts to using this user patch (used by humpman on personal-view)

Code:
[Patch8]
Name=2997i to 2400
Target=GH1 v1.32

E1_Offset=0x4014CB9A
E1_Check=0x0BB5
E1_Type=word
E1_Modify=0x0960

E2_Offset=0x4014D192
E2_Check=0x0BB5
E2_Type=word
E2_Modify=0x0960

[Patch9]
Name=2997i to 2397
Target=GH1 v1.32

E1_Offset=0x4014CB9A
E1_Check=0x0BB5
E1_Type=word
E1_Modify=0x095D

E2_Offset=0x4014D192
E2_Check=0x0BB5
E2_Type=word
E2_Modify=0x095D

Seems like enabling Patch8 (2997i to 2400), makes it more stable by not using the native 24p wrapper, but at 48i it can be easily conformed to a 24p timeline without any pulldown.

In such case, could the bitrate (in theory or otherwise) be increased with stability? (keeping the patch for 60 fps to 29.97 enabled of course, and changing 720p30 to capture at 720p24 to make it easier to conform to 24 timeline but with 4:2:2 color capability).

Ideally I'd like to push the 100mbps patch further if it can (taking from the 75mbps peak reliability the 720 resolution of 1536x864), make it 48i

The full user.ini humpman provided just in case one of the patches #1-7 has any impact on the last two

Code:
;Universal settings

[Patch1]
Name=Constant 512000
Target=GH1 v1.32

Input_Label=Input constant:
Input_Check_Min=0
Input_Check_Max=0xFFFFFFFF
Display_Offset=0x7BC20354
Display_Type=dword

;;Interlaced
E1_Offset=0x7BC20354
E1_Check=0x0007D000
E1_Type=dword
E1_Modify=input

E2_Offset=0x7BC2120E
E2_Check=0x0007D000
E2_Type=dword
E2_Modify=input

;;Progressive
E3_Offset=0x7BC270A0
E3_Check=0x0007D000
E3_Type=dword
E3_Modify=input

E4_Offset=0x7BC27E7D
E4_Check=0x0007D000
E4_Type=dword
E4_Modify=input

[Patch2]
Name=Constant 500
Target=GH1 v1.32

Input_Label=Input constant:
Input_Check_Min=0
Input_Check_Max=0xFFFF
Display_Offset=0x4014CAEA
Display_Type=word

E1_Offset=0x4014CAEA
E1_Check=0x01F4
E1_Type=word
E1_Modify=input

[Patch3]
Name=Constant 393216
Target=GH1 v1.32

Input_Label=Input constant:
Input_Check_Min=0
Input_Check_Max=0xFFFFFFFF
Display_Offset=0x4014C93D
Display_Type=dword

E1_Offset=0x4014CB00
E1_Check=0x00060000
E1_Type=dword
E1_Modify=input

[Patch4]
Name=Min QP
Target=GH1 v1.32

Input_Label=Input constant:
Input_Check_Min=0
Input_Check_Max=0x33
Display_Offset=0x7BC22E41
Display_Type=byte

;; Three ways together, both interlaced and progressive

;;Init QP, interlaced
E1_Offset=0x7BC22E41
E1_Check=0x14
E1_Type=byte
E1_Modify=0x33

E2_Offset=0x7BC2325C
E2_Check=0x14
E2_Type=byte
E2_Modify=input

E3_Offset=0x7BC23534
E3_Check=0x14
E3_Type=byte
E3_Modify=input

;;Init QP, progressive
E4_Offset=0x7BC297F2
E4_Check=0x14
E4_Type=byte
E4_Modify=input

E5_Offset=0x7BC29A29
E5_Check=0x14
E5_Type=byte
E5_Modify=input

E6_Offset=0x7BC29C6F
E6_Check=0x14
E6_Type=byte
E6_Modify=input

[Patch5]
Name=Max QP
Target=GH1 v1.32

Input_Label=Input constant:
Input_Check_Min=0
Input_Check_Max=0x33
Display_Offset=0x7BC23528
Display_Type=byte

;; Three ways together, both interlaced and progressive

;;Init QP, interlaced
E1_Offset=0x7BC22E21
E1_Check=0x33
E1_Type=byte
E1_Modify=0x33

E2_Offset=0x7BC23248
E2_Check=0x33
E2_Type=byte
E2_Modify=input

E3_Offset=0x7BC23528
E3_Check=0x33
E3_Type=byte
E3_Modify=input

;;Init QP, progressive
E4_Offset=0x7BC297EC
E4_Check=0x33
E4_Type=byte
E4_Modify=0x33

E5_Offset=0x7BC29A23
E5_Check=0x33
E5_Type=byte
E5_Modify=input

E6_Offset=0x7BC29C69
E6_Check=0x33
E6_Type=byte
E6_Modify=input

[Patch6]
Name=Scaling table for P-frames i
Target=GH1 v1.32

;;Interlaced
E1_Offset=0x7BD039A8
E1_Check=0x0C090807
E1_Type=dword
E1_Modify=array:0x0016001E001E00320032003200640064  0064006400960096009600C800C800C80016001E001E003200  320032006400640064006400960096009600C800C800C80016  001E001E003200320032006400640064006400960096009600  C800C800C8

[Patch7]
Name=Scaling table for I-frames i
Target=GH1 v1.32

;;Interlaced
E1_Offset=0x7BD03A68
E1_Check=0x0F070A05
E1_Type=dword
E1_Modify=array:0x30FF30FF30FF30FF30FF30FF30FF30FF  30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30  FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF  30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30FF30  FF30FF30FF

[Patch8]
Name=2997i to 2400
Target=GH1 v1.32

E1_Offset=0x4014CB9A
E1_Check=0x0BB5
E1_Type=word
E1_Modify=0x0960

E2_Offset=0x4014D192
E2_Check=0x0BB5
E2_Type=word
E2_Modify=0x0960

[Patch9]
Name=2997i to 2397
Target=GH1 v1.32

E1_Offset=0x4014CB9A
E1_Check=0x0BB5
E1_Type=word
E1_Modify=0x095D

E2_Offset=0x4014D192
E2_Check=0x0BB5
E2_Type=word
E2_Modify=0x095D

So far in testing just a minor changes from the base 100mbps patch

  • disabled 24p native wrapper
  • enabling 2997i to 2400 (60i to 48i) (in user.ini, patch #8)
  • checked 720p30 to 720p24 under Motion Jpeg
  • changed 720p30 resolution to 1536x864

With everything else the same I did some short non-scientific tests without any issues, basically under direct sunlight did bunch of paning shots against those pick-n-pluck foam with varying aperture between f/2 away from the sunlight to f/16 looking out the window and then focusing close to the foam which has a lot of tiny fine detail (closest thing to a foliage test I can think of) in both AVCHD and MotionJpeg. Didn't run into any issues or write issues (using a Promaster Rugged 32gb and 16gb card that writes at 80MB/sec and reads at 99MB/sec (which it does consistently when I tested with a dedicated USB 3.0 reader)

Seems a post from weimar over on personal-talk suggested being able to hit 120MB/s with audio rate to 320 if the native 24p wrapper were disabled with today's cards (his post was in 2013). Though I would probably 'choose' to reduce the audio rate since now days I primarily just use it to slate to an external recording (and being unable to play it back in camera wouldn't allow me to 'check' the in-camera audio out in the field)

But even if I don't get a response to this, I'm leaving this here for others to see, especially since the mere ability to keep it interlaced but at 48i seems pretty convenient.

Nutshell with the ability to conform to 24p with 48i (turning off the native wrapper), what bit rates can be pushed with some margin of safety?

Edit This is with the 75mbps patch, with the change set for 2997i to 2400 enabled (making the video 48i) and edited on a 24p timeline (after being transcoded to DNxHD).

 
When you are transcoding, what are you setting to make it 24p. Not sure why I'm asking as I'm selling my GH1, but just worth the info.

Somehow the interlaced recordings have a lower overall bit rate, even though they are twice the frame rate (24p vs. 48i), I still don't quite grasp this. Same goes for the ATSC 1080 30psf standard (60i vs. real 30p), it's really just the odd lines of a 30p frame in one field, and the even lines from that same frame in the second field. Bit rate should be the same but somehow this segmented frame allowed for broadcast where a real progressive frame did not.
 
When you are transcoding, what are you setting to make it 24p. Not sure why I'm asking as I'm selling my GH1, but just worth the info.

Somehow the interlaced recordings have a lower overall bit rate, even though they are twice the frame rate (24p vs. 48i), I still don't quite grasp this. Same goes for the ATSC 1080 30psf standard (60i vs. real 30p), it's really just the odd lines of a 30p frame in one field, and the even lines from that same frame in the second field. Bit rate should be the same but somehow this segmented frame allowed for broadcast where a real progressive frame did not.

I'm not entirely certain how to explain the latter but far as transcoding I'm just using Aiseesoft MTS converter and choosing 24fps as the target (it identifies the original MTS file as 24fps) into a DNxHD format.

I ran this GH13 SreamParser on the original file of the video here (since MediaInfo still identify it as 24fps, but also notes its interlaced)


FfnlWBh.png


With the information box containing :

Code:
File: B:\Videos\2017-07-13 Pens\PRIVATE\AVCHD\BDMV\STREAM\00000.MTS
    Size = 2,511,372,288 Bytes (13,080,064 Packets)
Stream: Max Speed = 220,595,700 bps
    Mode =  (actual frame rate is half)
    PCR Timing Interval = 0.095 seconds
    Clip Time = -7:-444:24780.00
    Average Total Bitrate = -742,460
    Average Video Bitrate = 548,693
    Average Audio Bitrate = -4,900
    Average Other Bitrate = -16,491
    Max GOP Size = 2,906,496 Bytes (23,251,968 bits)
    Video Frame Size (Min/Ave/Max)= 26,112 / 184,088 / 395,136
Frame: Video P Frame

Bitrate viewer for that video :

LpK3Swp.png


And MediaInfo states

Code:
General
ID                                       : 0 (0x0)
Complete name                            : B:\Videos\2017-07-13 Pens\PRIVATE\AVCHD\BDMV\STREAM\00000.MTS
Format                                   : BDAV
Format/Info                              : Blu-ray Video
File size                                : 2.34 GiB
Duration                                 : 9 min 24 s
Overall bit rate mode                    : Variable
Overall bit rate                         : 35.6 Mb/s
Maximum Overall bit rate                 : 120 Mb/s

Video
ID                                       : 4113 (0x1011)
Menu ID                                  : 1 (0x1)
Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : High@L4
Format settings, CABAC                   : No
Format settings, RefFrames               : 2 frames
Codec ID                                 : 27
Duration                                 : 9 min 23 s
Bit rate mode                            : Variable
Bit rate                                 : 34.0 Mb/s
Maximum bit rate                         : 16.6 Mb/s
Width                                    : 1 920 pixels
Height                                   : 1 080 pixels
Display aspect ratio                     : 16:9
Frame rate                               : 24.000 FPS
Standard                                 : NTSC
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Interlaced
Scan type, store method                  : Separated fields
Scan order                               : Top Field First
Bits/(Pixel*Frame)                       : 0.683
Stream size                              : 2.23 GiB (95%)
Color range                              : Limited

Audio
ID                                       : 4352 (0x1100)
Menu ID                                  : 1 (0x1)
Format                                   : AC-3
Format/Info                              : Audio Coding 3
Format settings, Endianness              : Big
Codec ID                                 : 129
Duration                                 : 9 min 24 s
Bit rate mode                            : Constant
Bit rate                                 : 192 kb/s
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 48.0 kHz
Frame rate                               : 31.250 FPS (1536 SPF)
Bit depth                                : 16 bits
Compression mode                         : Lossy
Stream size                              : 12.9 MiB (1%)
Service kind                             : Complete Main

So I'm not entirely sure how 48i is being woven or if it's basically 24psf (which splits exactly half of two frames into one frame at any given time and can be easily deinterlaced).

The way it was described by humpman when I asked the same on personal-view was

@kbeezie i don't clearly underastand your problem, but:
Stock GH1 NTSC mode is 24p and it's packed in 60i interlace with 3:2 pulldown video file.
With "2997i to 2400i" 24p will pack in to 48i interlace without 3:2 pulldown file.
Can't see problem here.
 
Hi, I just wanted to say thanks for the hack and the patches. I bought a second hand GH1 to play with it a bit and I'm enjoying experimenting with the different patches :)

I did a test just for fun, comparing my small GM1 with the GH1 updated with this patch in a static scene at ISO 1600, and the results were surprising (at least for me). The sensor on the GM1 is so much better at high ISOs, that the higher bitrate doesn't matter much (GM1 around 22000kbps, GH1 around 35000kbps).
On the GH1, Smooth film mode with everything on -2 except color. On the GM1, normal mode with everything on -5 except color.
It looks like the GH1 is slightly brighter at the same ISOs than the GM1.

A few crops below (GM1 left, GH1 right:
gm1_gh1_d.jpg

gm1_gh1_c.jpg

gm1_gh1_b.jpg

gm1_gh1_a.jpg

The GM1 crops more from the Panasonic 20mm 1.7 than the GH1 in video.

The videos:

 
It's been way too long, but the GH2 is the icon for these patches...

35 Mbps is kind of low. Once you start getting into the 150/175 Mbps with some of those final hacks from Vitaliy and Driftwood and company as those journey years came to an end, you'll see a nice improvement. (Some are more stable than others.)

And the GH1 and GM1 are 4 years apart! Sensors are super important for these results.
 
I think the 35Mpbs are due to the lack of details, but in this case I totally agree that the sensor is super important.
Even in the RAW photos there is no detail in the shadows, MFT sensors had evolved a lot since the GH1 :O
 
I recently bought a GH1 to play around with. Sometimes it's fun to go back to the older bodies and be reminded of what we had available to us a decade ago. It kind of puts into perspective what we have now.

The GH2 was my main shooter for 2-3 years, but I never owned a GH1 (I was in the 7D crowd back then). It feels like a homecoming of sorts. It was fun to re-live the excitement of hacking the camera (LPowell's Blackout). I could swear that the GH1 grip feels chunkier and better in my hand than the GH2, which was one of my big issues with that camera. No HDMI out is unfortunate and the dynamic range isn't great but I do generally like the image and the color palette though, as far as hybrid mirrorless cameras go. And adding a cheap Viltrox focal reducer makes it a lot more versatile (though the wider sensor does cause some of my APS-C lenses to vignette):

 
As someone still learning the ropes, I like seeing simple and yet competent videos like this by people who know what they’re doing. Little details like the sped-up speedboat. Portland looks like a pleasant place.
 
That's very kind but if I really knew what I was doing, I would have had the foresight to bring my whole filter pouch instead of just the one filter! ; )
 
Back
Top