RAW-video session coocked my sensor (the sad story of a stupid DP)

Started by adrjork, July 25, 2017, 07:24:43 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

adrjork

Please, I really need your advices, guys. The title of this story is "Red Dots FOR MY BAD".

This is the plot:
Yesterday I had my first real-&-long RAW-video session with 5D3 + Magic Lantern. It was early morning, I had no extra light, then I saw nothing via viewfinder (an HDMI external Ikan) so I decided to crank up the camera ISO to 12800 just to see the scene, finding the positions, the POVs, etc. (in video mode, 5D3 alerts you if you use ISO >1600 i.e. the white number turns yellow-bold on camera's screen, but I didn't know why). The temperature was very high 52-55° (again yellow-bold alert, but again I didn't know the consequences). I used in this way the camera for a couple of hours uninterruptedly (always ext. viewfinder on, always ISO 12800...) that is a long time (!), in fact I didn't use batteries but the DC-power-false-battery adapter.

Result: once transferred the video-tests on the computer, since the very first video I saw many red dots, in other words I cooked the sensor. In fact I made some tests (stills at low ISO with the lens cap on) and I recognized a lot of dead pixels (...a lot...) End of the story.

Questions:

1. In this very sad "Italian neorealism" story there are 3 components:
A) Super-high ISO;
B) Very long time with viewfinder opened (i.e. sensor exposed);
C) Too high camera's operating temperature.
The 1st question is: which of these components caused dead pixels? I can simply say that if the very first shot had red dots, this means that 12800 ISO is sufficient to cook the sensor. BUT at this point I wonder if the other two components could ALSO take part the sensor's damage? Is it possible in your opinion/experience?

2. You are experts of using 5D3+ML for RAW-video shooting, so my 2nd question is: which is the correct behavior of the DSLR video-operator for working with the camera without cooking its sensor? Well, obviously not cranking up ISO, now I understand that, but seriously which are your advices to use safely the 5D3 during a video production session? Perhaps using the camera's eyepiece all the test-time, and open the sensor (use the viewfinder) only for the definitive shot?

Thanks really a lot for your help (I need it, in these days of sorrow...)

Danne

Not really sure what is going on here. Can you provide sample MLV files?

Walter Schulz

Startup cam without loading ML (SET pressed), shot some stills in RAW and take a look if the sensor is cooked or not. I expect it will work just fine.

Levas

Just wondering, did you ever shoot mlv's with iso 12800 before?
Hot/dead pixels is very normal with that high iso's.

When you start up the camera and take a picture in low iso, for example iso 400, how does the picture look ? I expect normal pictures without red dots?

You can try remappjng dead/hot pixels by putting on the bodycap and do a manually activated sensor cleaning. Google for it, dead hot pixel remapping canon.

After this, some new high iso files probably have less red dots.

adrjork

Thank you for your replies.

I just took 4 black stills (cap on lens) WITHOUT Magic Lantern (using a blank SDHC formatted in-camera) at low ISO (tested 100, 800, 3200), and unfortunately the dead pixels are still there, even at 100 ISO...
I can count 10 dead pixels or so.
(If you really want to verify it, I can upload the 4 shots tomorrow because here in Italy uploading is very slow.)

Of course I can remap dead pixels via "manual cleaning sensor" method, but this method doesn't resuscitate dead pixels, simply it maps dead pixels to "mute" them and substitute them with interpolated values. It's a good trick (or a good palliative) to patch up a "troubled" sensor for continuing working.
Being under warranty, I'm thinking of asking for a sensor replacement, I'll see...

Anyway, I'd really like to understand how to avoid getting worse with my camera and to avoid cooking a possible new sensor (or maybe a future new 5D3). What is your work-method when you need to keep the camera working for a long time uninterruptedly? (i.e. when you need to see the scene through the camera for a long time uninterruptedly?)

(From your replies - thanks for your help - It seems that you don't think a very high ISO, or the high temperature, or the long time sensor exposure can damage a sensor, isn't it?)

Thanks really a lot.

Danne

There´s usually some hot/dead pixels on every sensor. I have a few as well on my 5D mark III. Are you sure they weren´t there before? Anyway. In this case uploads of example files says a lot more than descriptions.

Kharak

Have you tried Manual Sensor Clean in Canon Menu ? I got rid of some bad ones on my 5D2 back in the day.

I cannot remember the correct procedure, but go to Sensor Cleaning and set to Manual Sensor Cleaning and I think I recall you have to Push and hold or something like that for the duration or it starts and doesn't stop until you turn off your camera.

I read 30 Sec is enough, but I did 1 min to be sure. It might help.
once you go raw you never go back

g3gg0

i can ensure you, ML isn't causing any defects to your sensor.
we are letting do canon their stuff and just fetch the raw data on the software side.

if the things you did really caused some dead pixels to pop up during LV, they would have also appeared without ML.
Help us with datasheets - Help us with register dumps
magic lantern: 1Magic9991E1eWbGvrsx186GovYCXFbppY, server expenses: [email protected]
ONLY donate for things we have done, not for things you expect!

adrjork

Quote from: g3gg0 on July 26, 2017, 12:18:21 AM
i can ensure you, ML isn't causing any defects to your sensor. [...] if the things you did really caused some dead pixels to pop up during LV, they would have also appeared without ML.
Hey guys, I absolutely agree about that!!!
Let me explain better: I DON'T SAY that ML coocked my sensor! NOT AT ALL!!! (I love ML!) Instead, I'm saying that something in MY WAY of camera operating cooked my sensor (the title of this topic is "the sad story of a stupid DP"!)
I'm simply trying to understand if you think that a long-long time (2 hours uninterruptedly) of LiveView can damage the sensor. (Can it be?) Or if the super-high ISO can. (Can it be?) Or if the high camera's operating temperature can. (Again, can it be?)
These are three simple questions that have nothing to do directly with ML, but we know that ML-guys (you in this forum) install ML mainly for video, right? So, I think that you probably are video-oriented DPs or camera-operators, right? So, my question is simply: during your test-sessions - let image a long test-session in a set that you want to "explore" - do you usually use LiveView always on? For long time uninterruptedly? Without fear of damaging your sensor? OR you have a method (a way to operate, a trick, a tip...) to save your sensor?

That's the sense of my question.

Really thank you so much for your help.

rsmith02

I don't think that operating temperature is high for a Canon camera, and the camera would shut itself off if it were actually overheating. A few dead pixels out of a matrix of millions doesn't matter for anything. Remap and move on.

Walter Schulz

There are some ways to kill a sensor: Laser and ionizing radiation, brute force ...
I don't know Canon's policy on tolerated number of hotpixel calling for sensor replacement. But I'm quite convinced some 10 pixels won't trigger them.
We have cams with a ton of pixels intentionally not doing the job you are expecting: http://www.magiclantern.fm/forum/index.php?topic=16054.0

If there are issues like "cooking" with higher temperatures/long time use we should have heard about it. Lots of. Shitstorms and class suit stuff included. We haven't.

adrjork

Quote from: Walter Schulz on July 26, 2017, 07:38:08 AM
If there are issues like "cooking" with higher temperatures/long time use we should have heard about it. Lots of. Shitstorms and class suit stuff included. We haven't.
So, do you confirm that my way of operating (again, long time with LiveView ON, or with ext. viewfinder always on, and 52-55°) is a normal/common way of operating?

Thanks a lot.

Walter Schulz


a1ex

I've kept my 5D3 in LiveView for many hours, recording raw video over and over for testing purposes. Here's an example (30 hours, with external power supply). Don't remember the temperature, but it didn't trigger any warnings.

Want me to repeat the experiment and take some test images before and after?

adrjork

Quote from: a1ex on July 26, 2017, 11:18:44 AM
Want me to repeat the experiment and take some test images before and after?
You are so kind, but it's obviously not necessary: you've given to me the confirm that my way of using the camera (LiveView all the way) is common.
About camera warnings, it seems that in video-mode camera doesn't like ISO 3200 or more (so no warnings up to 1600). Also, I've noticed that higher the ISO value (and perhaps also larger the lens aperture), higher the operating temperature: this could be explained by the energy involved in ISO gain, and the amount of light (then heat) on the sensor at larger aperture. In fact, at ISO 100 and f/4.5 the operating temperature remains around 46-47°, while at ISO 400 and f/1.4 the operating temperature rapidly rises to 52-55° (yellow warning).

Just curious if someone can confirm (for experience) that 52° is common during shooting and safe for the sensor.

N.B. The temperature information I refer to is the one included in the ML overlays.

Thanks a lot

Datadogie

You say two hour video. I would have thought that you would progressively see it getting worse  as you watch the video. ie how does the first five minutes compare to the last five minutes?
T3i and Kiss X4 (550d (T2i)) Tamron 18-200mm, Sigma 28-70mm f2.8 (need firmware upgrade) Olympus 50mm f1.8  Olympus 28mm f2.8 and Olympus 24mm f2.8
Fancier 370 tripod and LCD hinged loupe. DIY Slider and crane.

adrjork

Quote from: Datadogie on July 26, 2017, 01:29:05 PM
how does the first five minutes compare to the last five minutes?
The dead pixels appears at the very first frame, and - it seems to me - they remains the same up to the last frame.
I can tell two things:
1. I took some stills the day before, and in those pics I can't see any dead pixel;
2. I had never tried ISO 12800 before.
Can it be a coincidence? It seems to me that damned ISO 12800 damaged the sensor burning some of its pixels...

Danne

You have to compare images from the same location. If you shoot a black frame it will reveal dead pixels while with normal shooting they most probably will not show. Do tests, normal shooting, normal iso, then shoot a black frame and compare...

adrjork

Done actually! As I said, now my black test reveals dead-pixels, and they are visible in the normal photos too. And I have an old black test that was perfect. Obviously I haven't a black test made the day before the "12800 day" (to prove that before the "12800 day" the sensor was perfect), but making a comparison between the normal photos taken the day before with those taken and the day after, it seems evident that dead-pixels appeared only from the 12800 shot (and now they are visible also at ISO 100 or so).
Anyway, again, I was simply curious about the DP common way of using the reflex for a video session, and your answers confirm that the LiveView always on, for long time, with high operating temperature, is all normal way of using the camera for video.
So, my theory is that something went wrong after gained the sensor at that level. Anyway I wrote to the Canon assistance, and we will see if they will decide for the sensor replacement or something else.

Anyway, really really thanks for all your replies and your help!

SpcCb

@adrjork > Note that the temperature indication on screen is not the sensor temperature, it's taken from the motherboard in the camera body. So high temperature number on screen, orange or red alert doesn't mean you are frying your sensor.

Since a couple of years I've done hardcore experimentations about temperature with 5D2 & 5D3 for scientific applications; taking long exposure photos up to 3h at high ISO (far worst to filming for the sensor), reach up to 80°C on the sensor (temperature record by a probe on the sensor, not the temperature on screen), also cooling sensors, up to -65°C (most of them pop off under -48°C), etc.

What I can say is that you probably will fry the motherboard _or the battery, with explosion risks_ before the sensor if you use your camera for a long time with liveview under very hot local conditions (I'm in Fr. so I know weather can be hot in Italia) and with Sun directly on it etc. The sensor is stronger than some other electronic parts in the body.

By the way it's possible to 'mark' the sensor and so get more hot pixels or noise after long time running under very hot local conditions, and also get a camera more sensible to thermal noise. Some kind of wear sensor level if you see what I mean. But like someone said, you can easily remove a couple of hot pixels, by remapping or in post if you filming in RAW.
If you want to see if your sensor is prematurely worn, put the camera in a fridge for 1/2h at 5~10°C (vegetables part) and then take a dark photo @ISO2500 with 1/50s exposure, you will see if it's not clean when camera is cold. If it's really a problem you can ask to Canon to replace the sensor, but don't tell them what you use your camera under very hot conditions because it's write to not do it in the manual.

Surin Dmitriy

I can confirm the same situation. I had long video session at Philippinnes (very hot conditions) and got dead pixels after that. So thats is sad true..
5D3

Surin Dmitriy

And by the way canon remapping algorithm doesnt work for such a case...I tried any combination and time periods but the hot pixels are still in place(
5D3

Walter Schulz

Dead pixel = Black all the time
Hot pixel = Max. level all the time.

What are we talking about here?

Surin Dmitriy

5D3

Walter Schulz

Canon's hotpixel remap function is known to have issues. Not working sometimes. You tried several times?