GH1 firmware research volunteers required

This is quite emotional report :-)

Lets make it step by step thing.
What you did on each step, camera state, freeze moment etc.
Check same thing having only incremented firmware.
Ok, less emotion next time :)
I will have to repeat the test step-by-step at home tomorrow to fulfil your request and will report back.
 
MJPEG M4 Test

MJPEG M4 Test

MJPEG M4 Adjustment Tests

All tests --> Apeture 4.0, Shutter 1/30, Manually white ballanced with on-camera light on.

MJPEG Quality Patched -> 384, 330, 300, 276
MJPEF Tables Patched -> 24, 24, 24, 24

Each adjustment tested in 4 conditions:

1) On-Camera Light On - ISO 320
2) Light On - ISO 1600
3) Light Off - ISO 1600
4) Light Off - ISO 100
-------------------------------------------------------------------------------------------------

Sensor m4 Adjustment 1 Tested with below settings

Set 1) 856 (Unpatched)
Set 2) 428 (Half of default integer)
Set 3) 214 (Quarter)
Set 4) 1284 (50% increase)

Results:

Yellow/Green color abnormality coming from bottom of screen in Set 2 and higher in set 3. Set 4 (higher setting) seems unchanged from unpatched. All changed seem completely static throughout each video frame. No other time/motion related changes noted in video.

Adjustment 1 Tests Stills --->> http://www.jerseyforged.com/sampleclips/m4adjust1_tests.zip

--------------------------------------------------------------------------------------------------

Sensor m4 Adjustment 2 Tested with below settings

Set 1) 435 (half of default integer)
Set 2) 1740 (double default integer)

Results:

Initial analysis of stills showed no signifcant change from base settings.
Same for video analysis of set 1 video showed no signifcant difference.
However, set 2's higher settings showed a pulse/flicker in the video. Initially not noticed at Lighted ISO 320 (but later found to be there to small extent), but very visible at ISO 1600 lighted and unlighted.

Videos (Rerendered/Edited for Time) of Unlighted ISO 1600 from Set 1 and 2 respectively ----> http://www.jerseyforged.com/sampleclips/m4_adjust2_test.zip


As mentioned, flickering is present but very very light at ISO 340 settings.. flickering may be directly related to ISO or simply less prevailent in darker scenes / lower exposures.


------- Update -------

M4 Adjustment 2: 1740 setting tested again at ISO 100 with camera lights and with below 1/30 shutters... no visible pulsating at all. Seemingly linked directly to ISO setting.
 
Last edited:
MJPEG m4 Adjustment 3 Test

MJPEG m4 Adjustment 3 Test

--------------------------------------------------------------------------------------------------

Sensor m4 Adjustment 3 Tested with below settings

1) 5 (Half Standard)
2) 20 (Double Standard)

Setting 2 (20) seems to have no noticable effect at all. Setting 1 (5) seemed to have none at first glance, but upon re-white balancing (as I've been doing before each test) I got this while balancing of a white sheet.....

P1000651.JPG


Upon setting any other non-manual white balancing mode, balancing functioned as normal. Manual white balance consistently gave this result. Setting 2 (20) seemed to have no effect on white balancing. May need to retest this however.
 
Last edited:
A reminder: no questions or requests in this thread, at all. This thread (and this whole subsection of the forum) is solely for development issues and reports from testers.
 
@GrgurMG

Are you shooting under fluorescent tube light? If so, the flickering might be due to interference between the frame rate of the camera and the flash rate of the tubes. Those light are not constantly on like incandescent lights, they flicker by nature.
 
@GrgurMG

Are you shooting under fluorescent tube light? If so, the flickering might be due to interference between the frame rate of the camera and the flash rate of the tubes. Those light are not constantly on like incandescent lights, they flicker by nature.

This setting can set some small delay so flickering become noticeable.
 
One strange request

Is it possible to test real resolution of MF assist mode using charts?
I want to check one theory but do not know how to do it.
 
One strange request

Is it possible to test real resolution of MF assist mode using charts?
I want to check one theory but do not know how to do it.
Not sure what you mean here, but I'll take a guess...

Are you asking if the resolved detail in the manual focus assist magnification is a lot sharper than the recorded frame? Because the answer is absolutely yes. What you see through focus assist is much, much sharper than the final 1920x1080 frame. And there is no aliasing or moire either.

I believe the manual focus assist is using a smaller window on the sensor, maybe even at pixel-to-pixel resolution.

The GH1 has an "extended optical zoom" mode where it windows off the sensor and uses a subset of the sensor, for very high quality cropped footage. Not a "digital zoom", it's a window of the sensor. Unfortunately that mode is only available in stills mode, not for video.

For video it has a "digital zoom" option, which is awful. At 2x and 4x it just scales the image up digitally, with all the loss of resolution that you would expect.

If we could get the "extended optical zoom" windowing mode in video recording, that would be fantastic because it would let us use all 16mm, Super16, and c-mount lenses without vignetting.
 
@GrgurMG

Are you shooting under fluorescent tube light? If so, the flickering might be due to interference between the frame rate of the camera and the flash rate of the tubes. Those light are not constantly on like incandescent lights, they flicker by nature.

Well the ambient room lighting is CFL and the on camera lighting is LED. Being related to the bulbs is a possibility as I think the effect might of been more pronounced in the background of the LED lit shots, however as Vitaliy said, there's something in the setting thats bringing it out. What is strange is that it only happens when you up the ISO, whereas it's non-existant at ISO 100. The flickering was noticable at what I think was 1/15 shutter and higher. Will have to run more tests later to be sure.

Furthermore, I'll run the m4 Adjustment 2 "5" setting test again later as well, as I'm not 100% convinced I might of somehow fubar'ed the results. I couldn't get any manual white balance abnormalities at the two other settings I tested.
 
Not sure what you mean here, but I'll take a guess...

Are you asking if the resolved detail in the manual focus assist magnification is a lot sharper than the recorded frame? Because the answer is absolutely yes. What you see through focus assist is much, much sharper than the final 1920x1080 frame. And there is no aliasing or moire either.

I believe the manual focus assist is using a smaller window on the sensor, maybe even at pixel-to-pixel resolution.

Here is that I mean.

On GF1 sensor setup sets two following parameters:
1) 2925 and 616 for MF assist mode.
2) 1394 and 1292 for MJPEG mode.
etc

If you take this two numbers and multiply on FPS you'll get same constant (54054000 in case of GF1 sensor).

First parameter is directly related to RAW width, as I think (it is 1920 or 2002 for GH1 FullHD modes).

So my question is - MF vertical resolution is much worse than horizontal one? From my observations it is not so.
Or this is really just window height that can be adjusted (seems reasonable if you remember that viewfinder resolution is 800x600).
 
Okay, I tested this by pointing at a 4K circular resolution chart, with a super-sharp lens in the sweet spot of the aperture range. The chart has res trumpets in a radial arrangement, so there are vertical, horizontal, and diagonal trumpets on it.

There are two magnification modes in the GH1, 5x and 10x. I checked both of them.

There is no way to record the results, of course, so the only way to validate what I was seeing was by relying on the LCD; the LCD may be a variable that throws the results off. Meaning, what if the LCD is sharper in one direction than the other? We may be measuring the LCD's performance, rather than the sensor's performance. But there's nothing I can do about it, I can only report the results.

Based on looking at the LCD at the 5x and 10x magnified images, it would appear that the image is substantially sharper/better resolved in vertical resolution, and less so in the horizontal. There is much more noticeable aliasing, at an earlier point on the chart, in the horizontal resolution (measured with the vertical lines). The horizontal lines (which measure vertical res) were cleaner and sharper with less aliasing.

I think that's backwards from what you were expecting to hear, but that's what I observed.
 
It is not backwards.
Windowed mode that uses 2925x616 stripe on sensor can have exactly this results.
This also gives simple answer to why we don't see cropped 1920 modes in any DSLRs.
But 640x480 mode in T2i is possible as it is central part of wide sensor stripe.
Souds reasonable?
 
I think I am seeing a problem in the native 24p patch. I wanted to do more testing before I presented my findings here and I have attempted to enlist help in the general sub-forum but I'm not getting a lot of interest it seems so I'll tell you what I'm seeing and what I'll be testing.

I've noticed that I have been able to run very high bitrates using both a class 6 card and class 10 card. These have both worked well for most tests.

What I have seen is that when I shoot a scene of highly detailed footage with moderate movement (hand held or moving scenery), I do not receive card speed errors regardless of which card I use or how long I record for. This works for bitrates I have tested up to 60+mbit/s.

If I shoot the scene from a tripod and the scenery is very still, I get card speed errors within 15 seconds of recording regardless of card used even for bitrates as low as 16mbit/s(stock settings).

This includes all of the "working" A,B,C settings in Ptool as well as the settings people have suggested.

The only thing different while using the stock settings was that Native 24p was checked. Once I unchecked it and retried, I had no issues.

I have now worked my way back up in bitrate and have been able to cleanly record 40+mbit/s to both my class 6 card and class10 card without Native 24p checked.

This leads me to believe that the overflow on the I frames could *possibly* be related to the native 24p patch.

I plan to do more testing to reach the limits of these cards without the native 24p setting and then supply a number of test files for analysis for I frame issues.

I'm simply not content buying a specific card to fix an issue that could be researched and fixed in a proper way.
 
Yeah, I apologize for posting in 3 threads. I guess I'm over zealous but I feel that there is a strong correlation here and while I understand people are trying to be helpful by suggesting I just need to buy specific cards, I believe that there is a problem that has possibly been identified that is being masked by simply buying cards that just happen to work good enough to overcome a problem. That's not a fix, that's a band-aid.

Hey I'm just trying to report on my testing.
 
Yeah, I apologize for posting in 3 threads. I guess I'm over zealous but I feel that there is a strong correlation here and while I understand people are trying to be helpful by suggesting I just need to buy specific cards, I believe that there is a problem that has possibly been identified that is being masked by simply buying cards that just happen to work good enough to overcome a problem. That's not a fix, that's a band-aid.

Hey I'm just trying to report on my testing.

Answer is very simple.
24p native patch changes GOP length from 15 to 12 (to make it standard 0,5sec).
If you want to record on same cards, just change GOP to larger values using appropriate patch.
 
Notice to G2 users.

We seem to be able to port most patches to G2.
But to keep it all pretty legal we must own G2 body (like we did in G1 and GF1 cases).
Special discounted body price for us is about $350+shipping.
So, if you want project to support G2, donate leaving note in PayPal form - "for G2 hack project" :-)
 
Notice to G2 users.

We seem to be able to port most patches to G2.
But to keep it all pretty legal we must own G2 body (like we did in G1 and GF1 cases).
Special discounted body price for us is about $350+shipping.
So, if you want project to support G2, donate leaving note in PayPal form - "for G2 hack project" :-)

does this means the G2 would work like a GH13 ?
will there be manual shutter speed ?

if not what can we expect from this G2 ?
 
Last edited:
does this means the G2 would work like a GH13 ?
will there be manual shutter speed ?

if not what can we expect from this G2 ?

Same as from GF1.
Manual shutter possibility is too early to tell (I am working on shutter and aperture related things, but it is in early state).
As soon as I'll have GF1 at hand I could tell about it in more detail.
 
Back
Top