E300 ISO 800 & 1600

[nokkieg]

I've read from the manual that ISO 800 and 1600 is "possible". Anyone care to explain what does this "possible" mean?

Does it mean some kind of in-camera digital manipulation to get ISO 1600 or is it something that has to do with pushing the exposure values (ie ISO 400 +2EV).

 

[Evilmerlin]

Yup, you need to enable 'ISO Boost' in the camera settings. I'm not particularly sure how it is accomplished though.

 

[ykkok]

The factory default has this ISO Boost turned off. So one will not be able to select ISO from 800 and above. Selectable values are 100, 200 and 400.

By turning this on, both 800 and 1600 are available. But please becareful and always be aware of your ISO setting for every shot you make. You'll not notice grain from the camera's LCD.

 

[nokkieg]

The factory default has this ISO Boost turned off. So one will not be able to select ISO from 800 and above. Selectable values are 100, 200 and 400.

By turning this on, both 800 and 1600 are available. But please becareful and always be aware of your ISO setting for every shot you make. You'll not notice grain from the camera's LCD.

Yes, shooting at high ISO, you will naturally get grainy pictures. This is sometimes desirable especially in b&w photography.

But I'm actually more puzzled by why Olympus has turned ISO boost off by default? Is it because the sensor can't actually handle the high iso or is it because the resulting grain is quite bad compared with other digital SLRs?

 

[tomcat]

Yes, shooting at high ISO, you will naturally get grainy pictures. This is sometimes desirable especially in b&w photography.

But I'm actually more puzzled by why Olympus has turned ISO boost off by default? Is it because the sensor can't actually handle the high iso or is it because the resulting grain is quite bad compared with other digital SLRs?

Olympus cameras are not the only ones that come out-of-the-box with the ISO boost turned off by default. The ISO boost in the Canon 10D and 20D is also turned off by default.

 

[mookie]

Yes, shooting at high ISO, you will naturally get grainy pictures. This is sometimes desirable especially in b&w photography.

Err..why is it desirable to get some grain in B&W photos?

 

[Merciless_Knight]

Yes, shooting at high ISO, you will naturally get grainy pictures. This is sometimes desirable especially in b&w photography.

But I'm actually more puzzled by why Olympus has turned ISO boost off by default? Is it because the sensor can't actually handle the high iso or is it because the resulting grain is quite bad compared with other digital SLRs?

Olympus cameras are not the only ones that come out-of-the-box with the ISO boost turned off by default. The ISO boost in the Canon 10D and 20D is also turned off by default.

To nokkieg;
Hint: You want Ur Cam's CCD lifespan to be shorted meh. :bsmilie:

 

[tao]

Err..why is it desirable to get some grain in B&W photos?

it is all about the expression of 'feeling' and 'mood'.

 

[nokkieg]

Err..why is it desirable to get some grain in B&W photos?

Hi mookie,

Looks like we're close nick relatives!

Well, sometimes in a b&w picture pictures, rougher grain is desirable because it adds to the tone and texture of the photograph.

Again, it varies with photo to photo. Sometimes in Portraiture, you don't want too much grain (eg bride in wedding gown). Sometimes you want a lot of grain (eg old man and the sea).

To me at least, grainy photos give a "natural matt" finish to it.

 

[nokkieg]

To nokkieg;
Hint: You want Ur Cam's CCD lifespan to be shorted meh. :bsmilie:

Hi Merciless Knight,

Thanks for the hint. Sorry I'm very clueless.

Will the CCD lifespan shorten if shooting at high iso? How so?

 

[imaginary_number]

I've read from the manual that ISO 800 and 1600 is "possible". Anyone care to explain what does this "possible" mean?

Does it mean some kind of in-camera digital manipulation to get ISO 1600 or is it something that has to do with pushing the exposure values (ie ISO 400 +2EV).

Usually, camera manufacturers term it as an ISO boost/expansion, which has to be enabled via custom functions/settings. This is due to the fact that at those sensitivities, the sensor's performance would deterioriate, e.g. reduced dynamic range or an exponential increase in noise. The E-300's case seems to be the latter, as seen in dpreview's noise test.

 

[ykkok]

All image acquisition device, CCD or CMOS have noise after some amplification or gain (increase ISO).

In camcorder world, increasing sensitivity is known as 'gain' and usually, 0dB gain is the cleanest and max +18dB is the noisiest.

For the E-300, I would say it's internal noise removal (or smoothing) engine is not as good as the others (or we could say that Olympus is trying to retain more details! may be) and let us have the flexibility to handle these noise in post-processing. But casual users may find ISO 800/1600 delivers shockingly high-noise pictures.

I've no problem with this since I've tried noise removal in post-processing on such ISO settings, and to my surprise, they are comparatively as good as other competitors!

 

[imaginary_number]

All image acquisition device, CCD or CMOS have noise after some amplification or gain (increase ISO).

In camcorder world, increasing sensitivity is known as 'gain' and usually, 0dB gain is the cleanest and max +18dB is the noisiest.

For the E-300, I would say it's internal noise removal (or smoothing) engine is not as good as the others (or we could say that Olympus is trying to retain more details! may be) and let us have the flexibility to handle these noise in post-processing. But casual users may find ISO 800/1600 delivers shockingly high-noise pictures.

I've no problem with this since I've tried noise removal in post-processing on such ISO settings, and to my surprise, they are comparatively as good as other competitors!

To be precise, noise removal can be implemented at the input stage or after the data has been captured, with the former being a better solution. Higher noise is not synonymous with increased detail. So the crux is not so much about which (post) noise-removal engine is better, but rather, the noise susceptibility of the sensor at the input stage, which is more of an instrumentation issue. In addition, with so many decent noise removal algorithms available these days, the bigger challenge among manufacturers in this arena is being able to get as clean a signal as possible at the input stage.

And to make the dB thingy clearer, gain is not always a good indication of noise level. The gain is fundamentally a measure of Vout/Vin. Noise is measured using Signal to Noise Ratio (SNR), which is average signal power/average noise power.

 

[ykkok]

To be precise, noise removal can be implemented at the input stage or after the data has been captured, with the former being a better solution. Higher noise is not synonymous with increased detail. So the crux is not so much about which (post) noise-removal engine is better, but rather, the noise susceptibility of the sensor at the input stage, which is more of an instrumentation issue. In addition, with so many decent noise removal algorithms available these days, the bigger challenge among manufacturers in this arena is being able to get as clean a signal as possible at the input stage.

And to make the dB thingy clearer, gain is not always a good indication of noise level. The gain is fundamentally a measure of Vout/Vin. Noise is measured using Signal to Noise Ratio (SNR), which is average signal power/average noise power.

You are right, higher noise is not synonymous with increased details, but I'm just referring to some people's pain when they complain that their ultra-smooth high ISO picture has some details lost, without letting them to control 'how much' to lose. Wouldn't it be nice to have noise removal level adjustable, such as the FZ-20's noise reduction function with 'Low-Normal-High' settings, instead of hard letting the camera decide?

And on the signal to noise ratio, is there any standard measures/tests.charts somewhere out there that compares each and every manufacturers' sensor?

 

[imaginary_number]

You are right, higher noise is not synonymous with increased details, but I'm just referring to some people's pain when they complain that their ultra-smooth high ISO picture has some details lost, without letting them to control 'how much' to lose. Wouldn't it be nice to have noise removal level adjustable, such as the FZ-20's noise reduction function with 'Low-Normal-High' settings, instead of hard letting the camera decide?

Well, that depends on which camera you're making references to - this thread's central issue is the E300's ISO boost. And the way I see it, this is not really an aggressive NR issue but plainly sensor characteristics.

Regarding the FZ-20 example, this is OT, but since u asked, my guess is that it was produced for a consumer market and it would be a more optimal solution (to Panasonic) to relieve the user of having to learn about NR, etc. Of 'coz, I won't deny that it'll be nice for the more advanced users (but this might not be the primary market segment they're looking at). Making a turn back to our topic, in the context of DSLRs, this is basically not an issue (RAW is an option).

And on the signal to noise ratio, is there any standard measures/tests.charts somewhere out there that compares each and every manufacturers' sensor?

Check out luminous-landscape.com, on Digital Camera Image Quality - not an extensive list, but something to get u started. Btw, if you've a good, discerning eye, as well as some decent knowledge, visual inspection at samples would be good enough.

 

[ykkok]

Well, that depends on which camera you're making references to - this thread's central issue is the E300's ISO boost. And the way I see it, this is not really an aggressive NR issue but plainly sensor characteristics.

Regarding the FZ-20 example, this is OT, but since u asked, my guess is that it was produced for a consumer market and it would be a more optimal solution (to Panasonic) to relieve the user of having to learn about NR, etc. Of 'coz, I won't deny that it'll be nice for the more advanced users (but this might not be the primary market segment they're looking at). Making a turn back to our topic, in the context of DSLRs, this is basically not an issue (RAW is an option).

Correct! I manage to get if not better, but as close to noise free (smooth) images from ISO 1600 RAW images in post-processing, comparable to the others.

Generally speaking, CMOS has higher noise (lower SNR) than CCD. May be some manufacturer have come out with better CMOS technology recently.

My humble opinion is, don't be afraid of experimenting with high ISO. Those nasty noise is removable. Silky smooth is easier to achieve than super sharp and detailed pictures, in terms of both resolution and dyn-range.

Check out luminous-landscape.com, on Digital Camera Image Quality - not an extensive list, but something to get u started. Btw, if you've a good, discerning eye, as well as some decent knowledge, visual inspection at samples would be good enough.

Thanks for the link, but I'm still hunting for a real comparison btw the benefit of having FFT vs IT CCD (ya OT here, sorry).

 

[imaginary_number]

Correct! I manage to get if not better, but as close to noise free (smooth) images from ISO 1600 RAW images in post-processing, comparable to the others.

Incorrect. Filtering noise inevitably removes details in an image which occur outside of the chosen cutoff frequencies - You won't get "better" images in comparison with one from a cleaner sensor. Close, maybe yes, maybe not - that depends on the image characteristics. "Comparable" or not depends on your threshold - if u do large printing for clients (and that's what I've done), then maybe not. If u do only small prints or downsample your images a fair bit, then obviously it doesn't really matter (to u).

Generally speaking, CMOS has higher noise (lower SNR) than CCD. May be some manufacturer have come out with better CMOS technology recently.

The 1st statement is controversial - and that's because of the continual efforts towards optimizing and improving on both technologies. More importantly, both have their strengths/weaknesses e.g. CMOS req. lesser power, resulting in lower thermal noise. So it wouldn't be fair to make an unsubstantiated general statement.

My humble opinion is, don't be afraid of experimenting with high ISO. Those nasty noise is removable. Silky smooth is easier to achieve than super sharp and detailed pictures, in terms of both resolution and dyn-range.

I never had any qualms over experimentation - by all means go ahead. But noise is not always removable - if you've studied instrumentation/digital comm., u'll understand why.

Thanks for the link, but I'm still hunting for a real comparison btw the benefit of having FFT vs IT CCD (ya OT here, sorry).

That's really OT and I'm glad u realized that. Since this is another subject altogether - start another thread to state your case if you're keen.

 

[ykkok]

Incorrect. Filtering noise inevitably removes details in an image which occur outside of the chosen cutoff frequencies - You won't get "better" images in comparison with one from a cleaner sensor. Close, maybe yes, maybe not - that depends on the image characteristics. "Comparable" or not depends on your threshold - if u do large printing for clients (and that's what I've done), then maybe not. If u do only small prints or downsample your images a fair bit, then obviously it doesn't really matter (to u).



The 1st statement is controversial - and that's because of the continual efforts towards optimizing and improving on both technologies. More importantly, both have their strengths/weaknesses e.g. CMOS req. lesser power, resulting in lower thermal noise. So it wouldn't be fair to make an unsubstantiated general statement.



I never had any qualms over experimentation - by all means go ahead. But noise is not always removable - if you've studied instrumentation/digital comm., u'll understand why.



That's really OT and I'm glad u realized that. Since this is another subject altogether - start another thread to state your case if you're keen.

You are right - filtering noise removes details, that's exactly what most of my Canon friends are complainint about.

Yes, CMOS generates less heat, so what's the advantage of CCD? And nobody has denied that Kodak CCD generates much higher noise, but that's not the end of the world. Most of these claims are scaring people away, and I personally don't think it's that serious.

By the way, have u got any chance to compare Olympus's RAW, before and after noise reduction, with the rests? I did and I've no complain here.

 

[imaginary_number]

You are right - filtering noise removes details, that's exactly what most of my Canon friends are complainint about.

Really? Maybe I'm ill-informed but I haven't had heard much of Canon (DSLR) owners who complained about the loss of details. More details (excuse the pun) would be good - and perhaps u could get them to post on CS their problems and we'll see where the issue lies. AFAIK, Canon doesn't use aggressive post NR (they've explicitly mentioned that their priority is noise minimalization at the i/p stage, go search Canon europe if interested), the comparisons made at dpreview's e300 review on detail resolution seems to confirm that as well. Additionally, the higher sensitivity performance was also documented on this page. If they're displeased with their Canon (and I presume u meant DSLR since u seem intelligent enough to be aware that it's the basis of this discussion), then may I ask u which other DSLR would provide better performance? (substantiated, please.)

Yes, CMOS generates less heat, so what's the advantage of CCD? And nobody has denied that Kodak CCD generates much higher noise, but that's not the end of the world. Most of these claims are scaring people away, and I personally don't think it's that serious.

I do know the advantages/disadvantages of the CCD/CMOs, made comparisons between them before jumping into the CMOs camp and if u're really interested yourself, then go do some reading up. And btw, I never said that high noise is the end of the world. You're right that many of these claims are scaring people away, but the fact remains that it matters for certain users who want maximal quality like myself. And like what I said previously, if you've a high threshold, it doesn't matter. Previously, when I was shooting casually, I used a G5 at ISO400 - noisy? Yes, but it doesn't matter in those cases.

Not serious for u maybe, but for people who are in the business, it can be serious. I had to help a friend DI a high detail image accidentally shot at 1600 for a seriously displeased client (not shot by a Canon, brand withheld for obvious reasons). For me, sometimes I like to crop a lot and play with around with exposure. With a noiser sensor (and I'm not making any references here), these luxuries would have to be done without.

By the way, have u got any chance to compare Olympus's RAW, before and after noise reduction, with the rests? I did and I've no complain here.

Good for u then. If u're happy and u know it, I'll clap my hands. :angel:

 

[hammer_400]

Just FYI....

from imaging resource...

on 1d mkII

....canon's cmos uniquely allows significant image-processing functions to be integrated right onto the sensor chip, before the image data even reaches the digitizing electronics. In the 1D Mark II's sensor, Canon has implemented no fewer than three different anti-noise processing techniques directly on-chip. While remaining very close-mouthed about the specifics of these anti-noise techniques, Canon does note that they affect both fixed-pattern and random noise generated within the sensor elements.

A common but little-recognized source of image noise in digital cameras is electrical noise generated by the digital signal-processing circuitry. The current spikes generated by high-speed digital circuitry can couple back into the sensitive analog circuitry, if circuit designers aren't careful. In the EOS-1D Mark II, Canon's engineers went to the extreme of completely separating analog and digital circuitry onto physically separate circuit boards, each with its own ground plane. It's hard to say just how much this reduced image noise levels, but it's notable that Canon invested the required engineering effort and manufacturing expense to achieve the required separation.....

On 20d

.... I suspect that this advanced noise reduction processing was another consequence of the "active pixel" CMOS technology Canon developed internally. Having active circuitry associated with each pixel in the sensor array allows lots of fancy processing that would be impossible otherwise, and it looks like Canon's noise reduction system takes advantage of this.

In the EOS-20D though, while apparently still using the sophisticated on-chip noise reduction processing we saw in the 10D, Canon has also added an option for conventional dark-frame subtraction as well. Accessed via Custom Function 02, the "Long exposure noise reduction" seems to operate just the same as dark-frame subtraction on other cameras we've seen it on. The difference with the 20D though, is that there's precious little image noise to be subtracted out, at least at exposure times of 30 seconds or less, where I did essentially all my shooting. I can imagine the dark-frame subtraction option being useful for astronomers doing 5-minute exposures with the 20D, but it will add little to most users' image quality.