SG9665GC High Contrast/Dynamic Range Flaw and other Discouragements

[Dashmellow]

most of the issue you have there relate to reflection, think we need to send you a CPL so you can do some comparison testing of these situations with/without

@jokiin, I don't mean to cause conflict or be adversarial but your post in response to @kamkar1's screen shots is another example of why I find myself becoming increasingly frustrated by my experience with the SG9665GC and your approach to this issue in particular after having spent a full year trying to deal with it and waiting for a remedy. In my initial post to this thread I used the term "sort of" in regard to your willingness to acknowledge the existence of the problems with the GC I've put forth here and this kind of reply is exactly what I was referring to.

Specifically what I said in post #1 was, "I've spent months working behind the scenes with @Pier28 and @jokiin attempting to analyze and resolve the problem........I've sent them most of the images I'm posting here and more along with other supporting data, including contrast curves and tone charts with the understanding from them that the problem is real and acknowledged (sort of) and that at some point it will hopefully be resolved with appropriate attention in the firmware."

When I first reported this issue to you and Pier28 a year ago it was immediately suggested to me that the problem was due to reflections and that a CPL would help or in fact actually fix it. After I finally established for you that the high contrast/high dynamic range issue was not due to reflections and not fixable with a CPL you shifted to suggesting that the problem was due to the dark hood of my vehicle in the image. Both privately months ago and again in this thread we have beat that horse to death several times over. The problem is not the dark hood in the FOV as I've clearly demonstrated. At one point in our private conversations you also suggested the problem "can come down to environmental influences", such as "a tree lined country road with bright sun in the middle".

So NOW, all of a sudden despite all the documentation, discussion, simultaneous comparisons with other cameras and repeated independent confirmations of the problem we are once again back to explaining it away to @kamkar1 as, "Most of the issue you have there relate to reflection, think we need to send you a CPL"

While @kamkar1's footage does start out with a bad windshield reflection that is CLEARLY not the thing he is trying to bring to our attention but rather to the EXACT same problem I have established.

Your general approach here seems to be to thank everyone for submitting videos and screen shots telling is how "helpful" this is, but at the same time basically dismissing the issue as a kind of "user error environmental thing". The fact is that the SG9665GC has a serious demonstrable "flaw" not seen in side by side comparisons with any other camera and I think you are in a certain amount of denial about it, at least publicly.

The road surface glare, blown out highlights and loss of upper tonal in kamkars images are surely NOT windshield reflections that can be fixed with a CPL.
These images have no dark hood in the FOV, nor a tree lined country road.
These images do demonstrate exactly the same tonal range/contrast/dynamic range issue I've tried to bring to everyone's attention and which feels as if it is not be adequately acknowledged.




Even when @kamkar1 turns the corner and the sun is now coming from a different direction, the white van is all blown out and is lost against the equally blown out white building.

 

[kamkar]

Without checking against a color scale i think both the white van and the building is RAL 9016 or 9003 in color, and thats about the whitest of white, RAL 9010 is allso white but a more creamy white.

It do appear as if the 9665GC is just as tight wound up in the daytime as it is for beeing able to make the night time footage it do, and i think it could maybe relax a little and that would provide a better result.
But its damm hard for me to say, i am used to have 3 knobs to tweak on my cameras, and i assume it is somthing else only having exposure and iso to work with.

I do think SG is hearing us and will be trying things to ease up the daytime footage, hell most other makes would just ignore us all together.

 

[jokiin]

He has large reflections across the image, your example washed out completely, his example doesn't come close to yours and as I said it would be good to see it with/without CPL to see how much is related to reflection, when he turns the corner the reflection is gone but the results are also far different as well

As I said previously we will be working on the IQ next month so collecting examples is helpful right now

 

[kamkar]

And i will keep providing footage, but i think i will let this thread rest now as what need to be said have been said.

 

[Dashmellow]

He has large reflections across the image, your example washed out completely, his example doesn't come close to yours and as I said it would be good to see it with/without CPL to see how much is related to reflection, when he turns the corner the reflection is gone but the results are also far different as well

As I said previously we will be working on the IQ next month so collecting examples is helpful right now

Whether one example is worse than any other one seems irrelevant. There is clearly a problem here and for me it is one I've been trying one way or another to deal with for a year. To repeatedly keep hearing about about dark vehicle hoods at the bottom of the images or window reflections after so much time is discouraging when I'm experiencing entire cars disappearing into the glare. I do appreciate the level of care an attention you give here but I was hoping for some more tangible effort to remedy the issue by now but it feels like there is somewhat of a reluctance to admit the actual problem. Based on some of your remarks I was waiting patiently for the forthcoming firmware update for several months now and was then quite surprised to learn that the IQ issues have yet to be addressed.

In the meantime, speaking of the "other discouragements" part of my thread title, I reported last week in the "SG9665GC Blurry" thread that my camera appeared to be going out of focus when parked on a mild day. I said I would keep an eye on it and today, a basically cool day with the ambient temps around 40-45 degrees Fahrenheit at the warmest part of the day and my camera apparently went seriously out of focus both when parked and while driving. It was cold enough that I was wearing a fleece vest and winter jacket. The interior of my vehicle was not in any way "hot". This camera is heavily promoted as being highly heat resistant and it is still only March in New England, not a heat wave in July. My last SG9665GC was extremely heat stable in all conditions but the following screen shots are from today. To repeat myself, I am not looking for conflict or wishing to be adversarial, I am just wanting a viable camera. I really don't know what to say at this point.

 

[alexp999]

The first thing someone on another forum said when I posted my footage was that I needed to turn HDR on even though it already is by default.

I've got a CPL and tipped the camera more towards the sky to try and help but it still frequently washes out and because it is pointing higher up it is more frequently looking at the sun.

I do love this camera for all its other benefits I just hope it's possible to tweak the AE so it is more consistent. I've held off reviewing the unit on Amazon because of this.

 

[jokiin]

We are still working on it, firmware development is ongoing as it always has been, further work on image quality is scheduled for next month

 

[THeBaNDiT]

One of my calibration attempts made my high contrast/dynamic range issue worse too!

Interesting story. As a photographer I did the calibration on one of my 5000K light tables but somehow the camera didn't like that very much probably due to the fluorescent tubes used in the light table. I redid the calibration again using the smart phone/iPad white image method and that was an improvement but not by much. That was on camera number two.

Maybe the dark pixel calibration is the problem. Yesterday I did bright pixel calibration and dark pixel calibration using white picture displayed by Samsung Galaxy S4 superamoled display on max - CAL status was OK, bright pixels still some star-like blinking. But it seems to show something like fog around bright places of picture. I tried bright pixel and dark pixel calibration using my tablet LCD and it seems to be better but not previously excelent.

Is it possible that dark pixel calibration will create the problem, if there is not proffesional 5000K light table? Can dark pixel calibration affect how the camera process the bright areas of picture?

 

[Dashmellow]

Maybe the dark pixel calibration is the problem. Yesterday I did bright pixel calibration and dark pixel calibration using white picture displayed by Samsung Galaxy S4 superamoled display on max - CAL status was OK, bright pixels still some star-like blinking. But it seems to show something like fog around bright places of picture. I tried bright pixel and dark pixel calibration using my tablet LCD and it seems to be better but not previously excelent.

Is it possible that dark pixel calibration will create the problem, if there is not proffesional 5000K light table? Can dark pixel calibration affect how the camera process the bright areas of picture?

I have mentioned several times in this thread including the above post you are replying to that I have tried calibration several times already. As also mentioned, I've now owned three different SG9665GC units. My most current version of the SG9665GC was shipped to me in late December, 2015. I was assured at the time that the latest iteration of the GC does not and will never require user calibration.

Importantly, @Pier28 has stated this publicly as well. Back on October 15, 2015 in the thread "A Few Issues", he said, "They locked the sensor calibration data into a protected memory area so it will no longer get over written during the next firmware update. Once the sensor is calibrated perfect at the factory, it will stay that way for the life of the product"

Also, another rational for not messing with the calibration on the most recent camera I received was so that I could continue testing the camera in different lighting and weather conditions (like after an awaited storm on a sunny day) and so I didn't want to change any variables on the camera itself so that I could try different aiming, CPL, dash mat, etc., etc.)

In my first post to this thread I went out of my way to make it clear to everyone that before posting I had tried every possible fix-it for this problem and that I've been dealing with the washed out, blown out footage for a year now. Please understand my frustration in fielding repeated "theories" and "explanations", etc., from well intentioned people who keep bringing up the same remedies and ideas for what is causing this that I made clear quite some time ago that I've already tried, especially the calibration thing at this point. I ask anyone reading this thread to go back and carefully review my first post.

What I said in my first post to this thread was,"..................the camera's performance is bad enough that it has had me contemplating taking it out of service completely. I say that as someone who has experienced this problem with three different SG units, different firmwares, calibrations, a dash mat, and two different CPLs."

For some reason, despite all the evidence so far including direct comparisons with other cameras in the same exact time and place there seems to be a reluctance among some to admit that for a number of us the camera has exhibited an inherent flaw. I want everyone to understand that I chose the word "flaw" to use in the title of this thread after very careful consideration in light of my experience with the GC. I do not use that term lightly since the idea of taking an expensive camera like this out of service because of such poor performance is something really I wish I didn't even have to consider after a year of dealing with the problem.

 

[jokiin]

Maybe the dark pixel calibration is the problem. Yesterday I did bright pixel calibration and dark pixel calibration using white picture displayed by Samsung Galaxy S4 superamoled display on max - CAL status was OK, bright pixels still some star-like blinking. But it seems to show something like fog around bright places of picture. I tried bright pixel and dark pixel calibration using my tablet LCD and it seems to be better but not previously excelent.

Is it possible that dark pixel calibration will create the problem, if there is not proffesional 5000K light table? Can dark pixel calibration affect how the camera process the bright areas of picture?

if calibration is not done correctly it can effect the picture, you can redo it if needed though

 

[jaro]

I was originally going to post this on another thread, but then I notice something that is seems relevant to this topic.
Not sure if this could be considered as an extreme condition. But the thing that causes some concern is the traffic light (the one above) is hardly seen at all up to the last seconds (but I guess I should be also glad at least the light was caught at all)

^ 'disaapearing building' is also a good title

direct link
http://www.mediafire.com/download/08k8azlskgh7pi3/MonkeySeeMonkeyDO.MOV

note:
GC is running on January beta and the exposure is set to +1 (iirc), due to my tinted windows.

 

[jokiin]

note:
GC is running on January beta and the exposure is set to +1 (iirc), due to my tinted windows.

thanks for the raw file, that helps

how dark is the tint on the window?

have you get any samples with EV at 0 for comparison?

 

[jaro]

thanks for the raw file, that helps

how dark is the tint on the window?

have you get any samples with EV at 0 for comparison?

Never did get the exact product details for the tint used in my car (1st car, newb to car at the time of purchase). All I know I have Medium 3M tint on the front and back and Dark 3M tint at the sides. (sales talk, I know )
I don't have any recordings that have EV at 0, will try to record this week (to help in the investigations), although not 100% sure that I could replicate the exact same scenario.
I did try setting the EV at 0, but night shots are affected and I reverted back to +1 (I usually drive when it is dark, above video is an exception).

 

[jokiin]

understand wanting to push the night results, +1 is a fair bit of adjustment though, this pics here will give you a bit of an idea about how much variation you get with each setting https://dashcamtalk.com/forum/threa...ev-exposure-value-settings.13732/#post-179228

 

[DT MI]

...note:
GC is running on January beta and the exposure is set to +1 (iirc), due to my tinted windows.

I would say that's not necessary at all. I have privacy glass in the rear of my vehicle with a CPL and only set mine to +2/3. My front GC with CPL and tinted glass is set to +0. I suspect you'd get much better results with it set to +0, or maybe even -1/3.

 

[alexp999]

I've tried dropping the angle of my cam back down and setting ev to -1/3. Will see how that goes maybe try -2/3 too but don't want to wreck night viewing. Hopefully a good comprimise while AE is looked in to as I get less compression artifacts with less sky in view.

 

[flip9]

Here is a short clip from mine, if it helps you diagnose the exposure issues.

 

[jokiin]

Here is a short clip from mine, if it helps you diagnose the exposure issues.

can you upload the original clip to a file host when you get a chance, that section of Canterbury road is good for this as it's mostly towards the east so gets good direct sun in the morning

 

[flip9]

That original file might have been overwritten on the card already.

I only have the fragment that i cropped using registration viewer. It seems to have retained the bitrates.

https://db.tt/2fS7YS4c

 

[jokiin]

That original file might have been overwritten on the card already.

I only have the fragment that i cropped using registration viewer. It seems to have retained the bitrates.

https://db.tt/2fS7YS4c

thanks, the fragment is fine, just needed the uncompressed version