SG9665GC High Contrast/Dynamic Range Flaw and other Discouragements

[reverend]

Have you ever done a manual sensor calibration on any of these @Dashmellow?

It's definitely worth getting the firmware updated - I went through 200GB of footage and nothing like what you saw here - I do have my cameras pointed further up though as I always find the Sony sensors like more sky in the image as they can overexpose things and clip where other sensors wouldn't. It's not an SG specific thing but more a Sony Exmor quirk from what I've seen.

 

[jokiin]

It's not an SG specific thing but more a Sony Exmor quirk in my experience.

the Sony sensor does tend to do this more than some other brands, it is something we're working on regardless though, no matter what there will always be room for improvement, no product is ever perfect for every situation, just need to keep at it with the incremental adjustments

 

[Dashmellow]

Have you ever done a manual sensor calibration on any of these @Dashmellow?

It's definitely worth getting the firmware updated - I went through 200GB of footage and nothing like what you saw here - I do have my cameras pointed further up though as I always find the Sony sensors like more sky in the image as they can overexpose things where other sensors wouldn't. It's not an SG specific thing but more a Sony Exmor quirk in my experience.

Yes, I've calibrated. I haven't re-calibrated this current one as I was assured by @Pier28 that this last one I received would not require any calibration as shipped. I've also not changed the firmware it was shipped with (12/15) either as I wanted to keep testing without changing any variables. As explained, we were waiting for a bright sunny snowy day but for the first time in memory here in New England that never quite happened this winter.

I'm going to try again with re-aiming the camera more skywards again though and see what happens. I don't know if it has had any effect on things but in the middle of all this I installed a very dark black dash mat, so that is a variable I hadn't thought of, I suppose. (It is a brushed suede mat that I spray painted with black camo paint and it works so well with killing windshield reflections I am truly astonished! Surprisingly the paint only improved it's handsome brushed suede appearance - much nicer than that carpet mat stuff!) Keep in mind that I've had three different SG9665GCs in hand and regardless of what I've tried I've had essentially the same result. Interestingly, the most recent camera has been an improvement over the first two. Camera number one had a focus issue along with everything else.)

 

[Dashmellow]

BTW, As previously explained, I was under the mistaken impression that the upcoming firmware update included IQ improvements but alas I'm informed we'll have to wait a bit on that. Still plan to try the update anyway.

 

[jokiin]

there are IQ changes in the 2016 firmware that differ to the earlier versions, those changes weren't specific to this issue but there is the possibility that they have some effect on the results so it is better to be using the newer versions for that reason, the IQ is something we have planned for more specific adjustment for this particular issue next month

 

[reverend]

@kamkar1 had one go tits up after the manual calibration which gave similar to what you saw in the videos:

Was that footage from the same camera?

It took me a few goes to get the manual calibration right - the automatic one from memory caused more problems than it fixed and was taken back out of the firmware - if your camera is running that version that may explain some of the problems maybe?

 

[jokiin]

I think in that instance @kamkar1 hadn't followed the calibration instructions to the letter, once he redid it he was back to normal

 

[Dashmellow]

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.

 

[Dashmellow]

It took me a few goes to get the manual calibration right - the automatic one from memory caused more problems than it fixed and was taken back out of the firmware - if your camera is running that version that may explain some of the problems maybe?

Which FW version are you referring to?

 

[jokiin]

Which FW version are you referring to?

it was in some of the private beta versions, you wouldn't have that version

 

[Dashmellow]

it was in some of the private beta versions, you wouldn't have that version

Oh. Thanks! Yeah, I've avoided beta versions for now considering the circumstances. Don't want to confuse the issue.

 

[jokiin]

Oh. Thanks! Yeah, I've avoided beta versions for now considering the circumstances. Don't want to confuse the issue.

that was prior to any public versions so not something you would have seen, the auto calibration worked fine but it broke something else in the process so we took it back out

 

[Dashmellow]

Oh, I see.

 

[James Kenitz]

BTW, As previously explained, I was under the mistaken impression that the upcoming firmware update included IQ improvements but alas I'm informed we'll have to wait a bit on that. Still plan to try the update anyway.

I'm really curious to see your results after trying the latest beta firmware and maybe angling up a click. IQ dropped noticeably for me when I experimented with angling down a click or two. I'm using the CPL filter with the thin mount (for now...thick is currently stuck to my windshield lol) and I get much better performance angled up to the max. When I dropped it down I would get some washout/contrast issues and it was generally poorer IQ all around. Not quite as bad as your situation but again it was noticeable.

 

[reverend]

Yes I'd definitely say get up to the latest firmware, do a full reset to defaults and then start logging data from there to see how things look - it could be a very specific set of circumstances triggering the problem in the older version that isn't present now.

And I'd definitely agree on moving it up a bit - the old rule of dashcam placement was from the days of Novatek 96650 and Aptina AR0330 and what works well for those doesn't necessarily work well for everything else and as with everything it's always trial and error to find what suits

The rear camera of my X2 I had to aim more like 70/30 sky/ road to get the exposure to play ball with my tints and rear shelf!

 

[Dashmellow]

I'm really curious to see your results after trying the latest beta firmware and maybe angling up a click. IQ dropped noticeably for me when I experimented with angling down a click or two. I'm using the CPL filter with the thin mount (for now...thick is currently stuck to my windshield lol) and I get much better performance angled up to the max. When I dropped it down I would get some washout/contrast issues and it was generally poorer IQ all around. Not quite as bad as your situation but again it was noticeable.

As I say, I will try. However, as I've pointed out several times the camera also has a tendency to underexpose in certain situations so tilting upwards has the potential to make that worse. I'll see what happens.

 

[jokiin]

the general rule of thumb is 40% sky, 60% road, with the Sony sensor though it does seem to fare better with more sky than what most sensors can, when there's a lot of hood in the frame even the colour of the car (dark or light) can make a difference, will be looking more closely at the AE table and how we can improve the metering to suit these situations

 

[Dashmellow]

Speaking of AE tables, least for me the camera(s) seem to have trouble reconciling extremes of contrast/brightness on both ends, so it could very well be influenced by my midnight blue truck.

@jokiin, do you remember when I sent you a screen shot that had been posted by another member who was experiencing the same thing driving on bright sand? He too had a dark hooded vehicle. Then again, there was so much bright area coverage in his image that it could be argued that tilting of the camera angle might have been irrelevant.

 

[jokiin]

we'll need to go through all these examples and see what adjustments we can make to the AE table that can help to deal with it the best

 

[reverend]

It would be interesting to see the AE tables - a7tool made it so easy with Ambarella!

Plus unlike your car @kamkar1 has a little girls car with hardly any hood showing in the front footage