Krontalk

Chronos => Chronos User Discussion => Topic started by: Nikon1 on February 04, 2021, 04:30:31 AM

Title: Influence of Sensor Temperature on Image Quality - Vertical Lines /Dynamic Range
Post by: Nikon1 on February 04, 2021, 04:30:31 AM
As kind of a follow-Up on this Thread:
 https://forum.krontech.ca/index.php?topic=608.0
 This New Topic here discusses the Impact/ Influence of Senor Temperature on Image Quality (on the 2.1, dont know how simmilar the 1.4 Sensor behaves about temperature, but if anyone is able to share any Information about it, its also very welcome here), And most importantly on the unpleasent Static Vertical Line Patterns/ Line Noise of the 2.1 in some situations (usually low Light or Stuff that needs to utilize the Full Dynamic Range of the sensor).
 The Old Thread is meant mainly to discuss the More Technical Aspects on how to achieve better Cooling than Stock on the Sensor to get those better Temperatures (Cooler Designs/ Cooling Methodes), while this one wants to talk about the Effects itself, that different Temperatures have on the Final Image.
Title: Re: Influence of Sensor Temperature on Image Quality - Vertical Lines /Dynamic Range
Post by: Nikon1 on February 04, 2021, 04:54:34 AM
First Results from the Watercooler Setup are in.
 Just tested 25,5C vs 45C Sensor Temp, while Lighting, Lens, Settings where all unchanged (cooling down and Heating up took a bit, so i must have knocked the Tripod a tiny bit, because of that image Shifted a little, sorry for that), For both Temperatures a Black Calibration was done when Temperature was very well Stable for a couple Minutes. Didnt do any Whitebalance, just had it on whatever Setting i used bevore, because I was Going to save in DNG.
 .
 Settings in detail:
 Lens: Jupiter-12 Lens 35mm f/2.8 set at f/4
 Camera: 1000fps; 1920x1080px; 0dB Analog+Digital; 180
 Sensor: Copper Waterblock With 11,4C Water for A Sensor Temperature of 25,5C and Without any Waterflow for a final Sensor Temp of 45,0C
 Lights: 2x Godox VL300, both at 85%
 Room Temperature was 20C This time
 .
 This Shot is pretty well Exposed, but you can still allready see quite a noticeable difference on the Vertical Line Patterns between them.
 Will Try to also do the Same Test again, but with a more Low Light Scenario, so it should be Way more Visible.
 I did actually expect there to be quite some amount of difference in Sensitivity/ Exposure between them also, but this seems not to be the Case, at least for 0dB and a well Exposed Image.
 .
 Had to pack the DNGs into a .zip, wouldnt let me upload them otherwise for some reason?
Title: Re: Influence of Sensor Temperature on Image Quality - Vertical Lines /Dynamic Range
Post by: Nikon1 on February 04, 2021, 06:57:36 AM
So here is an Low-Light/ Dynamic range Test. Settings overally are the Same as from the Test above, but Lights are now used with two Snoot Modifiers to generate a rather Sharp spot resulting in an very extreme Lighting Ratio/ High Dynamic range on the Final image, really pushing the ability of the sensor. Lamps Both turned up to 100% for this, Exposure Settings on the camera unchanged, as well as the Lens. Room Temperature was 19,5 to 19,0C this time, First "Cold" run was 10,3C Water temp (lower flowrate than last time to reach the same Temp as bevore), resulting in an 25,5C Sensor Temp again. "Hot" run was again without any Water Flowing, Sensor temp. while capturing the Footage was 43,0C (didnt wait forever to let it warm up this time, would propably still have gone up like 1C, because the Room Got colder meanwhile, but didnt want to wait even longer...Temp was fairly stable allready, so...). I did also include a DNG i recorded after it did heat up completely allready, but bevore Black-Calibration. Both of the other DNGs where shot after doing a Black-Calibration immediately bevore.
 .
 Take a look on the DNGs Yourself, if you want (Uploading them in a .zip again), but to me it seems like the Colder Sensor does a bit better in terms of Dynamic Range. Surprisingly, the Cold sensor seems to also have a little bit more of those Vertical lines visible in the Very, Very dark parts of the Frame, but on the Midtones and Highlights the Colder Sensor Footage looks way cleaner. This is however a very extreme Lighting scenario, and i wouldnt really want to push any Chronos footage so far, that the Lines in the Darker Parts of Frame would actually matter too much, but just wanted to point that out. The Midtones and Highlights, the parts of the Frame which usually matter the most in the Real world (Dark parts and Blacks would usually be crushed anyways in most cases, or if not Frame would propably be lit in a way where the "Blacks" are brighter than the Darker Parts in this frame here), look noticeable better when cooler. Not quite common practice in Filmmaking, as everybody would usually more tend towards underexposing (and boosting Shadows in post) than blowing out the Highlights, but what gives best results with the Chronos from what i have seen.
Title: Re: Influence of Sensor Temperature on Image Quality - Vertical Lines /Dynamic Range
Post by: Bian on February 04, 2021, 09:58:35 PM
The difference is hardly noticeable .....

From my experience the problem occurs at a lower aperture - its possible to try it with f2.8?
Title: Re: Influence of Sensor Temperature on Image Quality - Vertical Lines /Dynamic Range
Post by: Nikon1 on February 05, 2021, 01:06:45 AM
The difference is hardly noticeable .....

From my experience the problem occurs at a lower aperture - its possible to try it with f2.8?
All the Tests here where done on apperture Setting f/4 on an Lens that would be able to go to f/2.8 if i wanted to use it wide open. Both tests above where done with a Ton of light, two 300W LED Lamps iluminating a white surface of 22x15cm from about 30cm distance at basically full power. Given that i also "only" used 1kfps, 0db Gain and a reasonable Setting on the Lens, Both of those tests above are still about as ideal of lighting/ recording Conditions as i am able to create right now. So its trying to get no Noise at all if possible for some really high end Stuff. Basically there shouldnt be any noise to begin with, but there still is. So the differences between the both are sure subtle. Even the "Low Light Test" above was lit about as good as possible and should propably be called more of an Dynamic Range Test for now than anything else, if i think about it, even the Darkest parts of the Frame in that shot where propably still brighter than most surfaces in a Room with usual home Lighting...
 .
 While i am not really sure, what you mean with trying it at f/2.8 (do you refer to Less Light and Bigger apperture? if so, i can also go to f/1 and beyond if i am allready at it...), i want to do a ton more testing in more realistic Lighting Scenarios, test all sorts of gain/ Resolution/ Shutter angle settings, and even try some real Low Light / Low Framerate stuff (for available Light shooting in anything besides bright midday sunlight...). However, the Current setup for testing i have rn is VERY Sketchy at best, poses a serious Hazard to the Camera in multiple ways, is not portable at all, and takes a ton of time and work to setup and run tests on. All the Tests above where done in an Small Kitchen where i have a tap and a sink for the water, but not much space otherwise. Setup (mounting Cooler, setting up Lights, plumbing in Water, Tripods, lenses, Exposure...) takes currently 2+ Hours, and to run an single Hot-Cold cycle for a single test currently also takes 2+ Hours for Everything to get to temperature. I really would want to run so many more tests, but also only have so much time for this. Remember that after all of this, i still need to Remove the Whole Cooler assembly again and mount any other cooler Block on he Sensor without Tubing, and Clean up all the Equipment i used along the Way, if i happen to need the Camera for actually doing anything besides temperature-Testing, because it still has a bunch of Brass tubing sticking out, and a Hose attached, which will damage the Insides Badly if not handled 300% Carefully.
 .
 As an Conclusion so far, this was a very succesfull test/ Proof of Concept in my Opinion, showing, there IS in fact a difference in Image Quality with a Considerable cooler running sensor, so its worth it for me to actually go and figure out a way to make a improoved Cooling solution for everyday use.
Title: Re: Influence of Sensor Temperature on Image Quality - Vertical Lines /Dynamic Range
Post by: Bian on February 05, 2021, 02:34:02 AM
What i mean is, that there is not enough light, the aperture has to be completely opened (for example f2.8) and despite the open aperture, the maximum signal level is too low (<0,7V)
Title: Re: Influence of Sensor Temperature on Image Quality - Vertical Lines /Dynamic Range
Post by: Nikon1 on February 05, 2021, 03:25:41 AM
What i mean is, that there is not enough light, the aperture has to be completely opened (for example f2. 8) and despite the open aperture, the maximum signal level is too low (<0,7V)
Well, Blue Chanel is Allready Clipping on most parts of the Frame on the First one, Red and Green are Close to, and on the Second one, All Three Chanels are Clipping on some Pixels, so exposing any brighter than that, and it would just be Solid White Pixels on the Bright parts of the Frame. So, i dont see how to go much brighter without completely destroying the image by Overexposing it? I originally tried f/5.6 but that seemed actually too dark, went with f/4 and put the Lights down to 85%. Also tried f/2.8, but that was way too bright. f/2 or beyond wouldnt have left much of the Image.... 
 .
 I sure can and Propably will test completely overexposing eventually, but wasnt trying to do that here, as explained above, this was more about trying to minimize noise on an Well Exposed Frame to perfection, and pushing the Dynamic Range with tricky Lighting to see how the Sensor Handles that (in still close to ideal Conditions) at different Temperatures.