Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
01-11-2019, 11:25 AM - 4 Likes   #1
Pentaxian




Join Date: Dec 2011
Posts: 3,112
Mirrorless cameras and sensors - the good, the bad and the really ugly

This thread intends to spend a little time to analyze the many issues with sensors for mirrorless cameras and which problems and issues they suffer from - beyond typical DSLR sensor attributes.

Lets start with a discussion about inevitable sensor heat up and following image quality decrease through image noise increase.

Indeed, at least some earlier Sony models were known for not supporting extended video recording as they would shut down due to sensor overheating.

It is documented (The new Sony a7RIII camera still eats stars - Photo Rumors) that even the latest Sonys do eat stars due to their (it seems) constant noise reduction. You could call the Sony behaviour as "accelerator on steroids" as it obviously works all the time and not only from ISO 640 onwards.

QuoteQuote:
The smaller body design/size of the Sony A7RIII (and other A series cameras) results in a higher internal camera heat level which means in general a more aggressive noise reduction is necessary across the ISO range, but particularly important in the high ISO range and during long exposures because the accumulated effect of the activated sensor causes the heat to skew the signal/noise ratio and produces much noisier images. This is much more evident in higher resolution cameras like the A7R II & III because the more intense data processing generates more heat and the electronics of a higher resolution sensor means more dark noise in the signal/noise ratio. In order to resolve that a more aggressive noise reduction had to be applied at ISOs > 3200 which causes pixel level stars to be wiped. Pixel level doesn't mean actual pixel sized stars, the threshold is actually 4 pixels that make up an RGB block of the Bayer array since color info is needed to make noise reduction adjustment, so stars that are ~4 pixels or smaller are at risk of getting removed.
The phase detect AF sensors built into the imaging sensor generate dark noise in the sensor readout, and that is also especially noticeable at ISOs >3200. In addition, even though the PDAF system is not active during the actual capture of a photo, the electronics are passively in a "standby" mode and that generates additional dark noise. Because of this, the PDAF hot spot areas follow a noise reduction algorithm that is different than the rest of the sensor and accounts for that in order to deliver a smooth gradation and even noise pattern.
I was also told that currently because of the hardware limitations there is no way to resolve the issue and that it is not technically an issue at all, it is just a limitation of the hardware. Any fixes that would be applied via firmware would end up exposing the higher noise, which the engineering team determined would be a worse problem for users so it doesn't seem like a proper solution is in the works.
This does explain the suprisingly much inferior level of details you get to see from Sony cameras like the A7R3 when compared to Pentax and Nikon DSLR.

The Sony video overheating issue is still ongoing:
  1. SONY a7 III a73 OVERHEATING issue: Sony Alpha Full Frame E-mount Talk Forum: Digital Photography Review
  2. A7R III Overheat Issue: Sony Alpha Full Frame E-mount Talk Forum: Digital Photography Review
It seems there is no solution available even after years and many model changes.

I dont want to see full res Sony image noise under demanding low light long exposure settings after using video. You probably have to not use the camera for 15 minutes at least before you get back average noise /dynamic range performance from a glowing sensor.

I once measured a K-3 II and I wonder if someone has similar data available from a APSC or FF mirrorless camera.





You just need a camera, some time and an infrared camera to measure the sensor.

Maybe even metadata in the images documents temperature (often with 3rd party cameras it is body temperature thoughm which is much lower than sensor temperature).

01-11-2019, 01:09 PM   #2
Veteran Member




Join Date: Nov 2015
Location: California
Posts: 621
Interesting data, curious to follow this thread. Do you also have the environment data as far as temperature is concerned to go with this data?




---------- Post added 01-11-19 at 02:11 PM ----------

Oh and did you take the lens off for the temperature of sensor or are the temperatures with the lens on?
01-11-2019, 01:44 PM   #3
Pentaxian




Join Date: Oct 2010
Photos: Gallery | Albums
Posts: 6,241
this might be the reason why there is cooling fans in some of mirrorless video oriented cameras. It is interesting to see where this thread will go.
01-11-2019, 02:03 PM - 3 Likes   #4
Pentaxian
photoptimist's Avatar

Join Date: Jul 2016
Photos: Albums
Posts: 5,113
This is a fundamental drawback of an EVF -- the sensor runs full time in "video" mode. That 15° C higher temperature equates to 4X the dark current noise in the sensor.

01-11-2019, 02:35 PM   #5
Moderator
Not a Number's Avatar

Join Date: Mar 2012
Location: Venice, CA
Posts: 10,508
I wonder if the video mode sensor temperatures is part of the reason why Ricoh is reluctant to re-implement mechanical SR for video. Perhaps the noise problem they speak of isn't the mechanical noise of the SR mechanism being picked up by the audio but rather "noise" on the image.
01-11-2019, 10:19 PM   #6
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 11,250
QuoteOriginally posted by beholder3 Quote
It seems there is no solution available even after years and many model changes.
Well, a step towards the right direction would probably be to design the camera so that portability (compactness and lightness) are overriding priorities.
I believe -- without having anything concrete to go on but assumptions -- that it should be possible to achieve better cooling of the sensor by sacrificing some weight and size.

Sadly, one reads too little about the intrinsic downsides of mirrorless cameras and the solutions they often imply (e.g., OSPDAF).
One could forgive some of the press to gloss over some of the more technical downsides, if they didn't make a fuzz about issues that probably have a smaller visual effect.

QuoteOriginally posted by Not a Number Quote
I wonder if the video mode sensor temperatures is part of the reason why Ricoh is reluctant to re-implement mechanical SR for video.
Non-stabilised video operation should heat up the sensor just as much.

If Ricoh were concerned with image noise, they'd have to exclude or limit the use of video as such (including extended use of LiveView).

I suspect the real reason is that their sensor-based image stabilisation is optimised for still photography. For the latter scenario it is not important to achieve smooth transitions between corrective movements, on the contrary, jerky movements can help to increase the effectiveness of the mechanism. I think it is likely that if Ricoh wanted to support proper video stabilisation, they'd have to come up with a different stabilisation algorithm that yields good results during continuous operation.

I've read that Sony's video stabilisation doesn't (always?) yield pleasing results, so perhaps it is this kind of performance that Ricoh does not seem good enough for video stabilisation?
01-11-2019, 10:31 PM   #7
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 11,250
QuoteOriginally posted by photoptimist Quote
That 15° C higher temperature equates to 4X the dark current noise in the sensor.
Where did you find the 15°C figure?
The graphs show temperature deltas of 22°C and 7°C, I guess there is another source that I'm currently missing?

Where does the "+15°C -> 4X dark current noise" conclusion come from?

I'm not disputing anything, I'm just curious about the sources and the technical details.

Of course, the continuous sensor operation does not only heat up the latter but also drains the battery. On top of that, there is either a back LCD or an EVF to power even when the photographer just composes a shot or checks whether a scene is worth photographing.

I wonder whether some of the disparity between the low CIPA battery life figures for the Nikon Z6 (310) and the reports of much higher performance in the field (e.g. 600) are down to (strange?!?) shooting habits of people who claimed the much higher performance? Did they perhaps take every picture possible and did not spend much time on preparing the shot before pressing the shutter release button?

EDIT: The CIPA measurement procedure stipulates a 30s interval between images taken. During these "breaks", a MILC will typically consume more energy than a DSLR.

Sorry if this seems off-topic, but I think it is related in that it is something "bad" about MILCs and some of the lost battery life goes into heating up the sensor.


Last edited by Class A; 01-12-2019 at 02:04 AM.
01-11-2019, 11:50 PM   #8
CDW
Site Supporter




Join Date: May 2011
Location: Big Island, Hawaii & Utah
Posts: 457
I own a a7III (as well as a K1) and I've had absolutely no overheating problems shooting stills or 4K video. I don't shoot talking heads or do any blogging. My video clips are landscape/nature, usually no more than a minute long at the most. The video stabilization is effective but certainly not up to that of a steady cam.
01-12-2019, 12:04 AM - 1 Like   #9
Pentaxian




Join Date: Oct 2010
Photos: Gallery | Albums
Posts: 6,241
It is completelly different thing to shoot one minute clips than for example 5 minutes or more. If my memory serves me well it is even user manual that taking video or using LV for longer period could heat up the sensor(Pentax). So that is hardly surprising. But that it degreases performance. Little surprising, but really makes sense too..
01-12-2019, 02:03 AM   #10
Pentaxian




Join Date: Dec 2011
Posts: 3,112
Original Poster
QuoteOriginally posted by Fcsnt54 Quote
Interesting data, curious to follow this thread. Do you also have the environment data as far as temperature is concerned to go with this data?

---------- Post added 01-11-19 at 02:11 PM ----------

Oh and did you take the lens off for the temperature of sensor or are the temperatures with the lens on?
I don't think I have the exact environmental temperature data recorded. I think it was around 23 °C. I did quite some playing around before I did these recordings, so the camera body and sensors were already heated up to the levels shown.
I would expect that the increments will be even higher in the beginning if you start off in cool state.

Since I did measure the temperature with an IR camera it all was with lens off. I'd expect the more limited air circulation with a lens to cause even steeper increases.

---------- Post added 12th Jan 2019 at 10:19 ----------

Here is also more thorough information on Sony's star eater problem:

Star Eater: Documentation of an Issue with the Sony Cameras for Astrophotography (and How to Fix It) – Lonely Speck

and

Sony Star Eater

QuoteQuote:
The “Star Eater” problem is a form of software spatial filtering designed to reduce noise in photos, particularly hot pixels. Unfortunately, the rather rudimentary filtering algorithm that Sony is using easily mistakes sharp pinpoint stars for noise, nearly deleting them from the image or greatly reducing their brightness. The result is an astrophoto with less stars and the appearance of diminished resolution. Sony a7RII, a7SII and a7RIII cameras exhibit this problem for all exposure times longer than 3.2″.
and it was only partially fixed in firmware:

QuoteQuote:
The improvement to the "Star Eater" spatial filtering algorithm was made in firmware v4.0 for the Sony A7RII. It is believed it was also implemented in firmware v3.0 for the Sony A7SII but I haven't yet seen any proof.The improvement makes a difference only for the green pixels. The blue and red pixels are affected just as badly as previously. The effect is that more stars survive than previously but these "new" survivors are predominately green in colour.
One of the main weaknesses of consumer directed simplisitic websites such as DXomark and BClaff's blog pages are the poor documentation of the limitations of the chosen analysis/test setup, making them so untrustworthy.
In this case: Both completely close their eyes to / ignore poor camera performance in normal bulb mode for long term exposures.
If you look at their SNR curves, you'll find not even a hint at the massive raw data massaging going on at all ISO levels for long exposures.
01-12-2019, 03:57 AM   #11
Loyal Site Supporter
Loyal Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 27,601
For most applications, there probably isn't a big difference. Just as clearly, for all cameras, increase in temperature increases noise. I've seen this plenty of times with my SLRs. I've been shooting plenty of times on a summer evening where it is 30-ish degrees C and suddenly my K3 is producing all kinds of noise at iso 800. I'm guessing it probably does affect mirrorless cameras more and probably affects all video cameras more (I don't use my SLRs for video much at all).

I do think it is interesting that it seems that Sony is doing some aggressive stuff to long raw exposures on their cameras and aren't really being called out on it. I'm sure it is because when DXO Mark and Claff do their analysis it is on shorter exposures, but certainly it needs to be identified and they should be appropriately castigated for it.
01-12-2019, 09:21 AM - 2 Likes   #12
Pentaxian
photoptimist's Avatar

Join Date: Jul 2016
Photos: Albums
Posts: 5,113
QuoteOriginally posted by Class A Quote
Where did you find the 15°C figure?
The graphs show temperature deltas of 22°C and 7°C, I guess there is another source that I'm currently missing?

Where does the "+15°C -> 4X dark current noise" conclusion come from?

I'm not disputing anything, I'm just curious about the sources and the technical details.
22-7 = 15

In a silicon sensor, dark current doubles with each 8°C rise in temperature. Thus a sensor that is 16° hotter than another identical sensor will have 4X the dark current.


QuoteOriginally posted by Class A Quote
I wonder whether some of the disparity between the low CIPA battery life figures for the Nikon Z6 (310) and the reports of much higher performance in the field (e.g. 600) are down to (strange?!?) shooting habits of people who claimed the much higher performance? Did they perhaps take every picture possible and did not spend much time on preparing the shot before pressing the shutter release button?

EDIT: The CIPA measurement procedure stipulates a 30s interval between images taken. During these "breaks", a MILC will typically consume more energy than a DSLR.

Sorry if this seems off-topic, but I think it is related in that it is something "bad" about MILCs and some of the lost battery life goes into heating up the sensor.
You are right. The CIPA 30 second rule means a MILC uses most of its battery life idling rather than shooting.

Or, more importantly, the battery life of a MILC is best measured in minutes of on-time rather than shots. A person who quickly turns-on the camera, composes, and snaps a photo in 10 seconds might get 6 times the number of shots per battery charge as a person who uses a full 60 seconds for careful composing and waiting for the right moment.
01-12-2019, 01:33 PM - 3 Likes   #13
Pentaxian




Join Date: Dec 2011
Posts: 3,112
Original Poster
QuoteOriginally posted by photoptimist Quote
A person who quickly turns-on the camera, composes, and snaps a photo in 10 seconds might get 6 times the number of shots per battery charge as a person who uses a full 60 seconds for careful composing and waiting for the right moment.
It is casual holiday snapshots versus well composed shots or those for which you need to watch / follow / track the subject for a prolonged time (e.g. any wildlife, birding, sports). In the latter cases you will keep your eyes at the viewfinder for a long time per shot or series of shots.
This will require the DSLM to keep the sensor busy all the time and for long times.

If you just raise the camera to the eye every hour and then quickly take a shot that will not impact sensor temperatures in any ILC.

If I was a wildlife shooter and spending a lot of time checking the viewfinder at anyhow higher ISOs I'd be worried though about the impact of heat induced noise.
01-13-2019, 12:51 AM - 1 Like   #14
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 11,250
QuoteOriginally posted by photoptimist Quote
22-7 = 15
Ah, OK, I didn't look at this second-order temperature difference for a source of the 15°C figure for two reasons:
  1. The end temperatures on the graphs are not entirely comparable. If you take the 35min figure from the first graph, the difference only amounts to 12°C. I realise that the second graph appears to show an asymptotic behaviour so a temperature difference at the 48min mark would probably be close to 14°C or so, but it just didn't occur to me that you had extrapolated like that.
    .
  2. I don't think the 14°C (or so) temperature difference is that relevant for the subject at hand. First, the comparison here is between a DSLR in video mode and a DSLR in regular shooting mode. I'd be more interested in temperature differences between DSLRs and MILCs when using the same shooting modes. Second, I'm sceptical about a video mode reflecting a regular MILC usage well. I speculate that using "LiveView" on a DSLR would be fairer, as the electronics wouldn't have to work as hard and the battery would probably heat up less.
QuoteOriginally posted by photoptimist Quote
In a silicon sensor, dark current doubles with each 8°C rise in temperature.
Thanks. I did a search and figures seem to vary between 6-10°C for each doubling. Interesting to know, for sure.

QuoteOriginally posted by beholder3 Quote
If I was a wildlife shooter and spending a lot of time checking the viewfinder at anyhow higher ISOs I'd be worried though about the impact of heat induced noise.
For sure.

What would be really interesting, though, is a quantitative analysis regarding
  • how much a MILC (typically, or a particular one) heats up its sensor more than a DSLR in regular and/or wildlife shooting conditions.
  • at what point (e.g., regarding ISO settings) the respective increase in dark currents will visibly affect images.
If detrimental effects are only to be expected in fringe conditions then I don't think it would be fair to proclaim a general problem.

Full disclosure: I very much prefer optical viewfinders so I'm interested to learn about disadvantages of mirrorless systems (other than EVF issues) so that I can warn others that not everything will be greener at the MILC side. If almost everyone adopted MILC systems then an OVF may become a very niche feature. So I'm all in favour of identifying potential problems associated with MILCs, but I'd like to know whether the problems are realistically relevant. For instance, whether the sensor heat up is real (w.r.t. MILC vs DSLR) and then second, whether it is photographically relevant for non-extreme situations.

P.S.: Thanks a lot, @beholder3, for sharing your data here. I think it is a really useful start to the discussion.

Last edited by Class A; 01-13-2019 at 01:08 AM.
01-13-2019, 01:02 AM   #15
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 11,250
QuoteOriginally posted by photoptimist Quote
Or, more importantly, the battery life of a MILC is best measured in minutes of on-time rather than shots.
Interesting idea, but I'm not sure what is "best"; I guess it depends on the target audience.

I feel the CIPA conditions reflect serious photography better than a "many shots per on-time" approach does. A "on-time" could all too easily be converted into many images that wouldn't materialise in the context of serious photography.
As beholder3 put it, a snapshooter may get the so-called "real life" figures from a MILC, but a serious photographer might be closer to getting the CIPA figures.

I'm assuming that high-end cameras are more often used for serious photography and that testers do not always succeed in emulating the respective circumstances (to put it mildly) .

So comparing DSLRs with MILCS in terms of "minutes of on-time" would yield figures that would be useful for snapshooters (or technicians interested in efficiency) but wouldn't be as useful as an approach that considers that one can use a DSLR during "off-time" as well.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
banding, camera, cameras, data, eater, firmware, heat, iso, issue, k-1, lens, level, market, mirrorless, noise, overheating, pentax, photography, pixels, problem, reduction, sensor, sony, star, stars, temperature
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Photo Blogging - the Good, the Bad, & the Ugly R. Wethereyet Photographic Industry and Professionals 2 11-18-2018 10:15 AM
The Good - The Bad and The Ugly about Pentax kooks Pentax DSLR Discussion 41 10-28-2018 02:01 PM
Old used lenses: the good, the bad, and the ugly. catwalk Troubleshooting and Beginner Help 21 08-25-2018 07:26 AM
Architecture The Good, The Bad And The Ugly Kerrowdown Post Your Photos! 13 07-27-2018 10:49 PM
For Sale - Sold: The Good, The Bad and The Ugly - Film Cameras (US) Nowhere Matt Sold Items 3 05-27-2010 12:09 PM



All times are GMT -7. The time now is 01:02 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top