3K/UHD 5D2 Raw development and Other Digic IV Cams

[names_are_hard]

You're trying to call the function get_ms_clock_value(), but the compiler cannot find it.

You'll need to check whatever source you're using, but it may be that this function has been renamed to get_ms_clock().

[reddeercity]

You're trying to call the function get_ms_clock_value(), but the compiler cannot find it.

You'll need to check whatever source you're using, but it may be that this function has been renamed to get_ms_clock().

Thanks that help a lot , I'll check my source and see if this has happen .
 

[reddeercity]

This was a very learning experience! Show how thing are interconnected and simple change can made a string of thing get difficult 

You're trying to call the function get_ms_clock_value(), but the compiler cannot find it.

You'll need to check whatever source you're using, but it may be that this function has been renamed to get_ms_clock().

In short that's what i did and it complied , but to get to that point just what to explain
Would i had to do to understand what i was doing (its no good changing something if you don't know what go on , you need to fellow the bread crumbs even thou it may come to the same conclusion.)

So let dive a little deeper here , maybe this will help others or newbie/knobs to look in the code and make changes there's selves.
So to get started I when to the 1st line the error was "mj_rec.c:479:16:" and with help from @names_are_hard  about the function "get_ms_clock_value()" could not be found .
so i had to find where the "get_ms_clock_value" this was or a piece of it e.g. "get_ms_clock"

At the head of the file there's included other files to work in conjunction with the "c" to work.

Code Select Expand

#include <module.h>
#include <dryos.h>
#include <bmp.h>
#include <menu.h>
#include <config.h>
#include <property.h>
#include <raw.h>
#include <shoot.h>
#include <zebra.h>
#include <beep.h>
#include <lens.h>
#include <focus.h>
#include <string.h>
#include <battery.h>
#include <powersave.h>
#include "../lv_rec/lv_rec.h"
#include "../mlv_rec/mlv.h"
#include <patch.h>
#include "edmac.h"
#include "edmac-memcpy.h"
I had to compare each file we my newer source files until i found where the differences was.

Code Select Expand

/** timing */
/* todo: move to a separate file */
int get_seconds_clock();
int get_ms_clock_value();
uint64_t get_us_clock_value();
int get_ms_clock_value_fast();
int should_run_polling_action(int period_ms, int* last_updated_time);
void wait_till_next_second();Found it !! in DryOS.h , @Ant123 had a older version of ML and this was moved , now I had to find
where it was move to , (at this point getting scared this would mess everything up and would have to major work to get to compile    )
I look at my Dryos.h file and the the only difference was it had a file called "timmer.h"
Look in the source file and there it was 
there the "get_ms_clock"

Code Select Expand

int get_seconds_clock();
int get_ms_clock();    /* millis() on arduino */
uint64_t get_us_clock();
so now I'm confident that the chances will work because i found how this all connects

The Question is did all this work , kind of . The Module compiled , loaded on my 5D2 & 50D
didn't crash lockup , that good.
Menu loaded , enabled it was able to save a .avi file but there a problem with getting the video stream ,It looks like its hard code for the address on 450D trying to save a frame 848x560
see below , but not content .

Code Select Expand

General
Complete name                            : C:\Users\i5Dell\Desktop\MJpeg_Module_D4\MJREC.AVI
Format                                   : AVI
Format/Info                              : Audio Video Interleave
File size                                : 4.01 KiB
Writing application                      : Magic Lantern
Moti                                     : PEG Movie recorder module

Video
ID                                       : 0
Format                                   : JPEG
Codec ID                                 : MJPG
Width                                    : 848 pixels
Height                                   : 560 pixels
Display aspect ratio                     : 3:2
Color space                              : YUV
Compression mode                         : Lossy
So Ill need to refine this module , its pretty rough yet . it seem to be ok with crop_record module
so it need the address of the jpeg/hd buffer similar to mlv_rec with the raw buffer.

5D2


50D


mj_rec.cfg

Code Select Expand

# Config file for module mj_rec (MJ_REC.MO) mj.rec = 1so i know it loaded & (working ?)

Finally load the .avi in to an HeX editor and this is what i got
so its works or the .avi writer is work at least

Code Select Expand

RIFF....AVI LISTÀ...hdrlavih8...ÿÿÿÿ............................P...0...................
LISTt...strlstrh8...vidsmjpg............è....................'..............strf(...(...P...0.......MJPG.½..................
LIST....INFOISFT....Magic [email protected] JPEG Movie recorder module...............................
IC......2024-07-14 01:36:54 U.JUNK€............

...LIST....moviidx1....
Its really not too bad , going to need to do some work on the address thing , and hoping there's a way to tune the biterate also . 

[SoulState]

a fully funtional 1x3 crop_rec perset without adtg_gui module, the problem is the adtg reg are not applying for pixel binning "more then likely just a coding problem" I'm not a professorial coder , but i can get by.

Is there any progress with 1x3 crop_rec preset?
Honestly, this is the only/most awaited thing for 50d. Get rid of aliasing. 50d have great potential bitrate-wise, but aliasing issues ruins it all. It's just useless...

[Skinny]

Is there any progress with 1x3 crop_rec preset?

I actually tried activating 1x3 mode for 5D2 as it was described some pages ago in a video, and it worked... But it took me maybe 20 minutes or so to do it So it should be possible, in theory. And I think you will need a lot of mb/s by the way..

Congrats reddeercity, finding all this stuff - it looks really interesting Can't wait to try mjpeg recording

[SoulState]

I actually tried activating 1x3 mode for 5D2 as it was described some pages ago in a video, and it worked...

I tried this too, but for 50d with no luck, well, partially, posted image with artifacts (jagged edges) some pages ago.
Also no luck at all with 3.5k preset build (all pinky image), moreover, if i turn on fps override, and change fps from default, its getting worse - all screen and record with diagonal stripes. Some modules also conlficting in this build (links error or something)

[reddeercity]

I tried this too, but for 50d with no luck, well, partially, posted image with artifacts (jagged edges) some pages ago.
Also no luck at all with 3.5k preset build (all pinky image), moreover, if i turn on fps override, and change fps from default, its getting worse - all screen and record with diagonal stripes. Some modules also conlficting in this build (links error or something)

All crop_rec preset are tune to the framerate i set , it can not be changed !
its has to do with sensor's width x  height frame size limitations  plus Liveview,
certain frames size are tune either 24fps or 48fps(hi-speed framerate), there no offsets
to change framerate, it a whole kettle of fish to have offset framerates , the coding
would take a very long time (time i don't have) i have to distribute my time equally over the
things i need to finish (raw compression(mlv_lite)1x3 crop_record plus better preview image,etc....
i just don't have time for a new function.

I also notice the 50D crop_rec module is a little glitchy , if you have issue just clear your setting
specially  menu setting , the direct draw in liveview is not refreshing properly and some of the
setting don't show up in lower bottom of screen .
I'll also work on a updated crop_rec to fix some bugs to switching better between presets .
for the 50D. 
edit: i should be able to increase the frame rate on the hi-speed framerate from 48 tp 60fps and maybe a little bit more (70 maybe or even 75fps)
on;y reason why i can do this is because of the smaller sensor APS-C 1.6x crop factor

[reddeercity]

Looks like i my have found the correct reg's for CF card UDMA

Code Select Expand

CFATA] at CSMgrTas:FFB8BD98:FFB8BD94 [0xC0620202] <- 0x0       : ???
[CFATA] at CSMgrTas:FFB8BDB0:FFB8BDAC [0xC0620204] <- 0x0       : ???
[CFATA] at CSMgrTas:FFB8BDC8:FFB8BDC4 [0xC0620206] <- 0x0       : ???
Before we used the D5 5D3 base Reg

Code Select Expand

0xC062850Cbelow are configation for udma
https://www.magiclantern.fm/forum/index.php?msg=206010

Code Select Expand

+0x850C [32]  UDMA access timing
                 UDMA 0: 7D: 0x1213 5D3: 0x1001616
                 UDMA 1: 7D: 0x0C0C 5D3: 0x1000F0E
                 UDMA 2: 7D: 0x0909 5D3: 0x1000B0B
                 UDMA 3: 7D: 0x0607 5D3: 0x1000808
                 UDMA 4: 7D: 0x0404 5D3: 0x1000505
                 UDMA 5: 7D: 0x0203 5D3: 0x0000303
                 UDMA 6: 7D: 0x0202 5D3: 0x0000202I suspect the offsets to be the same , but the base is different
I found in a  FA_CaptureTestImage Log i had for while .

How do i know these are right , will after "0xC0620202"
the next bit of log show the successful configuration

Code Select Expand

[   CSMgrTask:ffb8bde0 ] (22:01) RequestConfiguration: SUCCESS
[   CSMgrTask:ffb8b958 ] (22:01) CF_DeviceCreate: StorageID=0, TotalBlks=0, HiddenBlk=0
[   CSMgrTask:ffb8b970 ] (22:01)                   WriteProtect=0, RotatingDevice=2
[   CSMgrTask:ffb89aec ] (22:01) cfSoftReset( 0 )But i don't think that its the only one needed to change configuration

Code Select Expand

[CFATA] at CSMgrTas:FFB8BD04:FFB8BD04 [0xC0628100] -> 0x0       : ???
[CFATA] at CSMgrTas:FFB8BD0C:FFB8BD04 [0xC0628100] <- 0x0       : ???
[CFATA] at CSMgrTas:FFB8BD18:FFB8BD18 [0xC0628100] <- 0x0       : ???
[CFATA] at CSMgrTas:FFB8BD24:FFB8BD24 [0xC0628100] -> 0x0       : ???
[CFATA] at CSMgrTas:FFB8BD2C:FFB8BD24 [0xC0628100] <- 0x2       : ???
[CFATA] at CSMgrTas:FFB8BD38:FFB8BD38 [0xC0628100] <- 0x2       : ???
[CFATA] at CSMgrTas:FFB8BD48:FFB8BD44 [0xC0628044] <- 0x0       : Interrupt related?
[CFATA] at CSMgrTas:FFB8BD5C:FFB8BD54 [0xC0620200] <- 0x41      : ???
[CFATA] at CSMgrTas:FFB8BD68:FFB8BD68 [0xC0620200] -> 0x0       : ???i think we may need these also , but i need to check farther ,
the first one to change is the 0xC062020x and see what happen
I try and test this tomorrow and see if this correct.

Edit: here the Log file
https://bitbucket.org/reddeercity/crop_rec_5d2_50d/downloads/FA_CaptureTestImage.txt

[reddeercity]

Tried the CF card Reg's , did see a little increase but not what i'm expecting
Basically i just recorded the 3X crop (5x zoom) @ 2152x1074 @23.976fps 14bit (default bit rate)
which requires 95MB/s i do believe , i was pecking as high as 85MB/s , average was 76.73MB/s which is good. I'm looking for at least 95MB/s
(i don't thing D4 will ever see 130MB/s , but 95 to 110MB/s seems achievable) there a lot of overhead on 5D2 as much as 20MB/s to 30MB/s
i have old documents that show this from a1ex back 5-6 years ago(i need to search for it) Normal Average write on 5D2 is 66MB/s to 68MB/s with MLV_Rec ,
the old first gen raw record max speed was 73MB/s (all @ 14bit)So if finding the correct reg's for UDMA7 seems out of reach  then i can start looking at reducing the overhead .
And i have gone down the path of overclocking the CF controller yet , i would prefer to easily set the UMDA by a simple over ride of regs .
With 0xc0620202

Code Select Expand

# Config file for module mlv_rec (MLV_REC.MO)
mlv.video.enabled = 1
mlv.res.x = 13
mlv.write_speed = 7673
mlv.preview = 2
mlv.display_rec_info = 2
mlv.buffer_fill_method = 0
Had pecking around 83MB/s

Code Select Expand

# Config file for module mlv_rec (MLV_REC.MO)
mlv.video.enabled = 1
mlv.res_x = 14
mlv.write_speed = 6644
mlv.display_rec_info = 2
mlv.warm_up = 2
mlv.buffer_fill_method = 0Without any CF Card controllerreg's applied

Code Select Expand

# Config file for module raw_rec (RAW_REC.MO)
raw.video.enabled = 1
raw.res.x = 14
raw.write.speed = 7318
raw.preview = 1
raw.warm.up = 2
raw.bpp = 12OLd Raw_Record (it was the fastest to date)
and again this Average write speed not peck or idle write speed.
So if i can get a 10MB/s average increase in write speed with one reg (guessing/iuck )
after some more investigation hopefully i can more   

[SebastianC]

This is good news! I hope fast speed Improves

[2blackbar]

10bit even 12bit  is possible with jpegs so...

Saving MJPEG frames in 10-bit color depth requires a few considerations and a customized approach, as standard MJPEG implementations typically use 8-bit color depth. Here are some steps to achieve this:

1. Understand the Requirements
10-bit Color Depth: Each channel (Red, Green, Blue) will use 10 bits, leading to higher color precision.
File Format: MJPEG (Motion JPEG) typically saves each frame as a JPEG image within a container like AVI or MOV.
JPEG2000: Unlike the standard JPEG, JPEG2000 supports higher bit depths, including 10-bit per channel.
12-bit JPEG: Some extended JPEG formats can support higher bit depths.

[Walter Schulz]

In ML supported cams video encoding happens in a black box (aka: We have no idea how to tweak it) and rn looks like hardcoded. There is not enough procession power in Digic to just add different realtime H.264 or JPEG2000 encoders via software.

[reddeercity]

Have a look this jpcore that what we have to work with.
Walter right , D4 is limited in CPU processing power , that's why there's encoder chips
on the main board ,the only way to get a customs codec is to re-flash the rom with the codec code.
not a option , but the JPcore chip does has pretty good compression.
Thur i know it have the ability 4.2.2 but it can do RGB(4.4.4) whether that 8bit or 16bit
but once again as Walter say's "its a block box" , So i'm tied to the canon's routine's
I do belive thou the MPU has a lot to do with the JPcore also , if i could understand the MPU string code there use , (its numbers & letters) i could understand the jpcore more.   

[reddeercity]

Look like i found the total CF Card routine in a startup log for my 5D2
Seems to have the "DesiredTuple" configuration I'm looking for
now to figure out what & where the offsets are set by which Reg's
 Seems 0x41 is the offset , or is it for UDMA 6.0 maybe 0x51 or 0x14 is UDMA 7.0 ?
Anyways i'll do some testing tomorrow
 

Code Select Expand

CSMgrTask:22:03: ---- CFEventHandler(ID=0:Event=8) ----
        CSMgrTask:22:01: CF_ResetCard SUCCESS
        CSMgrTask:22:01: SearchTuple: DesiredTuple=0x22, Next=0
        CSMgrTask:22:01: SearchTuple: SUCCESS
        CSMgrTask:22:01:              Offset=0x7e, Code=0x22, Link=0x2
        CSMgrTask:22:01: GetTupleData: DesiredTuple=0x22
        CSMgrTask:22:01:               Offset=0x7e, TupleDataMax=0x100
        CSMgrTask:22:01: SearchTuple: DesiredTuple=0x22, Next=1
        CSMgrTask:22:01: SearchTuple: SUCCESS
        CSMgrTask:22:01:              Offset=0x86, Code=0x22, Link=0x3
        CSMgrTask:22:01: GetTupleData: DesiredTuple=0x22
        CSMgrTask:22:01:               Offset=0x86, TupleDataMax=0x100
        CSMgrTask:22:01: SearchTuple: DesiredTuple=0x15, Next=0
        CSMgrTask:22:01: SearchTuple: SUCCESS
        CSMgrTask:22:01:              Offset=0x2c, Code=0x15, Link=0x23
        CSMgrTask:22:01: GetTupleData: DesiredTuple=0x15
        CSMgrTask:22:01:               Offset=0x2c, TupleDataMax=0x100
        CSMgrTask:22:05: [CIS] MakerInfo  : KINGSTON
        CSMgrTask:22:05: [CIS] VersionInfo: ULTIMATE CF CARD 16G
        CSMgrTask:22:01: SearchTuple: DesiredTuple=0x1a, Next=0
        CSMgrTask:22:01: SearchTuple: SUCCESS
        CSMgrTask:22:01:              Offset=0x90, Code=0x1a, Link=0x5
        CSMgrTask:22:01: GetTupleData: DesiredTuple=0x1a
        CSMgrTask:22:01:               Offset=0x90, TupleDataMax=0x100
        CSMgrTask:22:01: RequestConfiguration: pLStorage=0x684eac
        CSMgrTask:22:01:                       ConfigBase=0x200, StatusReg=0x0
        CSMgrTask:22:01:                       PinRepReg=0x0, CopyReg=0x0
        CSMgrTask:22:01:                       ConfigIndex=0x1, Present=0xf
        CSMgrTask:22:01: RequestConfiguration: Base = 200, Data = 41
        CSMgrTask:22:01: RequestConfiguration: SUCCESS
        CSMgrTask:22:01: CF_RequestConfiguration: err=0x0
        CSMgrTask:22:01: ConfigIndex = 1 Match
        CSMgrTask:22:01: CF_DeviceCreate: StorageID=0, TotalBlks=0, HiddenBlk=0
        CSMgrTask:22:01:                   WriteProtect=0, RotatingDevice=0
        CSMgrTask:22:01: cfSoftReset( 0 )
        CSMgrTask:22:01: cfSoftReset SUCCESS
        CSMgrTask:22:01: &CurCardInfo=0x4c9f0
        CSMgrTask:22:03: Word164 = 0x8d9b
        CSMgrTask:22:05: [ID:Firmware Revision] = Ver6.02K
        CSMgrTask:22:05: [ID:Model Number] = ULTIMATE CF CARD 16GB                   
        CSMgrTask:22:05: IDE = 4, PCMCIA = 80, UDMA = 6
        CSMgrTask:22:03: Cyl=16383, Hds=16, Trk=63, nSec=31227840, Ms=1
        CSMgrTask:22:01: cfDecideTiming: UDMA Mode 6 (CFA4.0)
        CSMgrTask:22:01: cfDecideTiming: UDMA Mode 6 (CFA4.0)
        CSMgrTask:22:03: CF_GetAccessTiming : DatTim = 3, DatMod = 6
        CSMgrTask:22:05: cfIdentifyDrive: Cache Support
        CSMgrTask:22:05: cfIdentifyDrive: Idle Command(500mSec)
        CSMgrTask:22:05: cfIdentifyDrive: Set UDMA( Mode=6 )
        CSMgrTask:22:01: pBlkDev=0x685200
        CSMgrTask:22:03: nBlocks=31227840, bytesPerBlk=512
        CSMgrTask:22:03: blksPerTrack=63, nHeads=16
        CSMgrTask:21:01: [fsuDeviceAdd] (0)
        CSMgrTask:22:01: cfReadBlk: st=0, num=1, ofst=0
        CSMgrTask:21:01: fsuGetPart: default 1
        CSMgrTask:21:01:             nBlocks = 31227840, blkOffset = 63


[reddeercity]

Did some test on the CF card controller reg's , What i though was the right reg
turn out it was wrong , the log wasn't correct . But i did find the reg that controls the
the cf bus speed , i can control it from 82.3 MB/s down to 5.4 MB/s ,
that right i said 82.3MB/s to 5.4 MB/s !
normally with get 68 MB/s to 72MB/s in some cases up to 75MB/s ( that's the average)
I'm about to get 82.3MB/s (i suspect this maybe UDMA 7 cause 75MB/s is the default now & that UDMA 6)
There a lot of overhead here , i think i'm going to poke at the reg & see if i can reduce the overhead
by turning off certain ones i've been researching if its necessary or not.

back to the Reg i found to work , i just used the last 4 digic from the 5D3 reg's
the 0x1 seems the fastest so far , i'm still poking this reg 
 

Here a list of what of the offsets that gear down the CF card controller bus.
tested 3x crop default mode (2150x1076 @23.976fps at 14bit)
had a 78 frame buffer

Code Select Expand

0x1 = 78-81MB/s
0x2 = 68-71MB/s
0x3 = 57MB/s
0x4 = 49MB/s
0x5 = 42MB/s
0x6 = 38MB/s
0x7 = 33.9MB/s
0x8 = 30MB/s
0x9 = 25.6MB/s
0xa = 25MB/s
0xb = 23.4MB/s
0xc = 22MB/s
0xd = 20MB/s
0xe = 19MB/s
0xf = 18MB/s
0x10 = 17MB/s
0x11 = 16MB/s
0x12 = 15.5MB/s
0x13 = 14.6MB/s
ox14 = 14MB/s
0x15 = 12.9 - 13.4MB/s
0x16 = 13MB/s
0x17 = 12.4MB/s
0x18 = 11.9MB/s
0x1a = 11.1MB/s
0x1b = 10.7MB/s
0x1c = 10.4MB/s
0x1d = 10.1MB/s
0x1e = 9.8MB/s
0x1f = 9.4MB/s
0x20 = 8.9MB/s
0x21 = 8.8MB/s
0x22 = 8.7MB/s
0x23 = 8.2MB/s
0x24 = 8.0MB/s
0x25 = 7.6MB/s
0x26 = 7.8MB/s
0x27 = 7.0MB/s
0x28 = 7.3MB/s
0x29 = 7.1MB/s
0x2a = 6.9MB/s
0x2b = 6.8MB/s
0x2c = 6.6MB/s
0x2d = 6.5MB/s
0x2e = 6.4MB/s
0x2f = 6.2MB/s
0x30 = 6.0MB/s
0x31 = 6.1MB/s
0x35 = 5.6MB/s
0x38 = 5.4MB/s

Some other i poked
0x1021 = 18MB/s
0x1022 = 12MB/s
0x121 = 80MB/s
0x1201 = 82.3MB/sbefore i thought 0xC0620200 was correct , to control the card bus ,
that's wrong,  0xc062850c is correct at least right now , unless i found something different.

[reddeercity]

poking the 50D for the same reg

Code Select Expand

0xc062850c Its the  same as the 5D2 ,

Code Select Expand

50D results
0x1 = 78-83MB/s
0x3 = 55
0x5 = 43.3
0x11 = 16MB/s
0x21 = 8.8MB/s
0x1302 = 25.5MB/s
0x1402 = 78-83MB/s
0x2402 = 15MB/s
0x1801 = 78-83MB/s
0x1807 = 18MB/sMore or less the same , maybe the 40D has the same Reg , i know it suffers from low CF bus speed

 

[reddeercity]

Looks like i may have found the lossless routine for 5d2 or should i say the missing parts

Code Select Expand

14.509.246  ShootSsDev:ff162f04:a7:03: sssExecuteMem1ToRawPath(26599)
14.509.308  ShootSsDev:ff162f50:a7:02: LOSSLESS_FLAG
14.509.386  ShootSsDev:ff163324:a7:03: ProcessTwoInTwoOutLosslessPath(R) Start(26599)
14.509.395  ShootSsDev:ff163330:MMIO : [0xC022200C] <- 0x00000012
14.509.419  ShootSsDev:ff3dd2c0:00:00: *** LockEngineResources(6830d8) x16:
14.509.440  ShootSsDev:000b06c8:00:00:      1)    10000 (read channel 0x8)
14.509.453  ShootSsDev:000b06c8:00:00:      2)        d (write channel 0x16)
14.509.466  ShootSsDev:000b06c8:00:00:      3)        2 (write channel 0x2)
14.509.480  ShootSsDev:000b06c8:00:00:      4)    30001 (read connection 0x1)
14.509.495  ShootSsDev:000b06c8:00:00:      5)    2002d (write connection 0x2d)
14.509.511  ShootSsDev:000b06c8:00:00:      6)    20016 (write connection 0x16)
14.509.520  ShootSsDev:000b06c8:00:00:      7)    50034 (?)
14.509.529  ShootSsDev:000b06c8:00:00:      8)    5002d (?)
14.509.539  ShootSsDev:000b06c8:00:00:      9)    50010 (?)
14.509.548  ShootSsDev:000b06c8:00:00:     10)    90001 (?)
14.509.559  ShootSsDev:000b06c8:00:00:     11)   230000 (?)
14.509.569  ShootSsDev:000b06c8:00:00:     12)   160000 (?)
14.509.580  ShootSsDev:000b06c8:00:00:     13)   260000 (?)
14.509.591  ShootSsDev:000b06c8:00:00:     14)   260001 (?)
14.509.601  ShootSsDev:000b06c8:00:00:     15)   260002 (?)
14.509.612  ShootSsDev:000b06c8:00:00:     16)   260003 (?)

14.633.967  ShootSsDev:ff3dc84c:00:00: *** UnLockEngineResources(6830d8) x16:
14.633.987  ShootSsDev:000b06c8:00:00:      1)    10000 (read channel 0x8)
14.634.001  ShootSsDev:000b06c8:00:00:      2)        d (write channel 0x16)
14.634.013  ShootSsDev:000b06c8:00:00:      3)        2 (write channel 0x2)
14.634.028  ShootSsDev:000b06c8:00:00:      4)    30001 (read connection 0x1)
14.634.043  ShootSsDev:000b06c8:00:00:      5)    2002d (write connection 0x2d)
14.634.059  ShootSsDev:000b06c8:00:00:      6)    20016 (write connection 0x16)
14.634.069  ShootSsDev:000b06c8:00:00:      7)    50034 (?)
14.634.078  ShootSsDev:000b06c8:00:00:      8)    5002d (?)
14.634.087  ShootSsDev:000b06c8:00:00:      9)    50010 (?)
14.634.097  ShootSsDev:000b06c8:00:00:     10)    90001 (?)
14.634.107  ShootSsDev:000b06c8:00:00:     11)   230000 (?)
14.634.118  ShootSsDev:000b06c8:00:00:     12)   160000 (?)
14.634.129  ShootSsDev:000b06c8:00:00:     13)   260000 (?)
14.634.139  ShootSsDev:000b06c8:00:00:     14)   260001 (?)
14.634.150  ShootSsDev:000b06c8:00:00:     15)   260002 (?)
14.634.161  ShootSsDev:000b06c8:00:00:     16)   260003 (?)

different terminology between D4 & D5 cams

Code Select Expand

sssExecuteMem1ToRawPath

Code Select Expand

LOSSLESS_FLAG
the other part , "TTJ" stuff missing in my code plus the "sss state"

Code Select Expand

14.645.411  ShootSsDev:000178e4:00:00: *** SSSState: (9) --10--> (2)          ff1639d8 (x=65ffcc z=65ffcc t=0)
14.645.436  ShootSsDev:ff163a14:a7:03: sssCompleteRawToLcdJpeg(26599)
14.645.456  ShootSsDev:ff163844:a7:03: sssStopMem1ToJpegPath(26599)
14.645.471  ShootSsDev:ff290cb0:MMIO : [0xC0F26408] -> 0x01596EE0
14.645.472  ShootSsDev:ff290cb0:MMIO : [0xC0F26508] -> 0x03201008
14.645.495  ShootSsDev:ff3cf300:16:03: [TTJ] STOP WR1:0x3201008 WR2:0x1596ee0
14.645.499  ShootSsDev:ff290cb0:MMIO : [0xC0F26A08] -> 0x0124E704
14.645.501  ShootSsDev:ff290cb0:MMIO : [0xC0F04A08] -> 0x0ACF938E
14.645.521  ShootSsDev:ff3cf32c:16:03: [TTJ] STOP RD1:0xacf938e RD2:0x124e704

Code Select Expand

14.659.528  RearShtDev:ff20d524:8f:03: RemoveObject : SRAW_OBWB not found
14.659.545  RearShtDev:ff20d544:8f:03: RemoveObject : LRAW_YUV not found
14.659.554  RearShtDev:ff20d564:8f:03: RemoveObject : MRAW_YUV not found
14.659.562  RearShtDev:ff20d584:8f:03: RemoveObject : SRAW_YUV not found
14.659.572  RearShtDev:ff20d5a4:8f:03: RemoveObject : JPEG_LF_YUV not found
14.659.582  RearShtDev:ff20d5c4:8f:03: RemoveObject : JPEG_LN_YUV not found
14.659.591  RearShtDev:ff20d5e4:8f:03: RemoveObject : JPEG_MF_YUV not found
14.659.599  RearShtDev:ff20d604:8f:03: RemoveObject : JPEG_MN_YUV not found
14.659.611  RearShtDev:ff20d9dc:8f:03: RemoveObject : JPEG_SF_YUV not found
14.659.621  RearShtDev:ff20d9fc:8f:03: RemoveObject : JPEG_SN_YUV not found
14.659.629  RearShtDev:ff20da1c:8f:03: RemoveObject : JPEG_S2_YUV not found
14.659.638  RearShtDev:ff20da3c:8f:03: RemoveObject : JPEG_S3_YUV not found
Jpeg Compression Levels ?, seems to right compaired to my Canon SDK kit

Code Select Expand

namespace Canon.Eos.Framework
{
    public enum EosCompressLevel : byte
    {
        Unknown = 0xF,
        JpegUncompressed = 0,
        JpegCompression1 = 1,
        Normal = 2,
        Fine = 3,
        Lossless = 4,
        SuperFine = 5,
        JpegCompression6 = 6,
        JpegCompression7 = 7,
        JpegCompression8 = 8,
        JpegCompression9 = 9,
        JpegCompression10 = 10,
    }
}So maybe the Black Box can be opened finally   

Plus I've taken a Deep Drive in to the Live View World , more precisely the HDMI in 3x3
looking into removing the Side Black bars , so the HDMI should go from 1650x1080 to 1920x1080 16x9 (no black bars)
I've come close , but haven't found the regs to open up the Horizontal , found the Vertical ones , but that's not what i'm after.
Don't expect anything for a while , being working on this for a week already and just scratched the surface plus there's many rabbit hole to fall in to.
 

[whitelight]

Reddeercity, this is great, great news. Thanks for your enduring efforts.

[reddeercity]

Small update about a few things i'm working on , some interesting results with some possibility's
1St, Liveview (HD Vram) looking for a reg or Reg's to turn 3:2 A.R off to 16x9 (tell you latter why i think its a matter of turning off a switch (code wise or reg maybe), the other thing i may need a bigger HD Vram buffer , just using the canon default one , which could be why my results are limited.
This was with the HDMI connected to my led monitor @ 60i in 3x3 mode full frame

I can move the whole HD liveview to the Left fully , (when recording raw video with the HD liveview like this doesn't effect the raw image it stay the centered etc. ...)



Moved over to the Right fully , same as above no problem recording raw video , raw image untouched by
HD Vram reg's.

Found this to be very interesting that need more in depth investigation , i found what the real
width of the HD vram is , (this was when i was recording H264)
if i can found the right routines i can get HDMI to match 


Lastly a simple quick bench mark for the CF card with the modified reg for max write speed
or as match as the overhead will allow , it more then normal , kind of impressive.
Usually don't hit 105MB/s in the bench mark , in photo mode by the way

 

[reddeercity]

Looks like i'm making some progress
Didn't know that the 5d2(D4) has 720p 59.94 fps HDMI output !
That I2c HDMI chip has a few possibility once we know have to configure it correctly.
So it appears i found full width HDMI in 3x3 mode but in 1280x720p 59.94 fps
it seem to fuul the full width , just need to adjust the verital offset & the Hozonal offset.
I have found the Offset also but didn't have time today to contine to refine it .
below are 2 Video links , one is the HDMI Capture thought my Aja Kona PciE card on my PC
Captured to 10 bit ProRes 422 proxy (53 Mb/s) then compressed it to MP4 720p 59.94fps (5Mb/s)
the other video clip is from my iPhone (4K to 1080p30) to show it working.


 5D2_HDMI_Full_with_720p59.94_fps.MP4 HDMI Capture file 60MB


iPhone video iPhone file 28 MB

First i put the HDMI in to progressive scan (1920x1080p 29.97fps)

Code Select Expand

c0f14254 0x32090231 ---> 0x22083232The Next Regs take some time to fine (and that was pure luck by the way)

Code Select Expand

c0f14210 0x277A ---> 0x277Bthat change to 720p 59.94fps , by the way i can do 720p50 108025i but i had no image there.
the very strange thing also i found the spot where i was getting 1080p59.94 but it was coming in and out with no real image


 

[nikki]

5D2 great...!

[reddeercity]

Seem that 720p50 work good , (it was by total luck that i got this to work this well in a short time)
Here the Regs i used

Code Select Expand

c0f14210 0x277a --> 0x277f
c0f14220 0xe1 --> 0x15
c0f14254 0x32090231 --> 0x2208fc01

I crop the bottom to 1280x535 then i vertcally sterched to 1080 , that ended up as 1280x1080 50FPS
4x3 Aspect Ratio , that telling me the HD buffer need to modified just like raw in crop_record
anyways i think what happening is the HDMI chip is resampling the 1650x1080(HD Buffer) to 1280x720p50
Its clean , definitely progressive scan and at 50fps, it also works in default Crop_Mode(2150x1074)
More work is needed , but i good start. I guess it could a good focusing aid .
   

[reddeercity]

just a litle update , it looks like all Reg's with

Code Select Expand

C0f14xxx Control the total HDMI screen which would be the HD Vram (if i'm not mistaken) the total width including blackbar the on the right & left
So far

Code Select Expand

C0f140xx to C0F142xxSeem to have a Horizontal distortion or skew from left leaning to right leaning.

Code Select Expand

C0F11308 does a Vertical Stretch , seems like its changing the pixel from square (1:1) to a 1:2 or 1:1.5
maybe enough to stretch the 720p50 down close to full vertical.

But i'm really looking for is to change the Aspect Ratio from 3:2 to 16:9
i see it the logs , the base Aspect Ratio is 3:2 (because its a Photo cam first , Video Second)
At least as far as the HDMI is concern , Liveview is 3:2 or 16:9 depending if in photo or video mode.
If i can find it then i would be portable across the ml D4 & D5 (since there are close at least the 5D2 & 5d3 are.)

[reddeercity]

Found the Vertical Stretch & Compressed in 3x3 , should be very useful in the HDMI development , maybe a module even who knows 
 

Also found a HDMI Regs that give a Multi View , pretty interesting maybe different preview for 10x ?
not too sure but it goes to show you how powerful the HDMI chip really is , now to tap in to that.