a900 ISO: 100 vs 320

[douglasf13]

Since ISO values are a bit abitrary in all cameras anyway, maybe Sony just needs a firmware release that adjusts ISO 200 to ISO 320, so that people get over the mental hurdle of using 320? BTW, this isn't unique to the A900. There are other cameras that exhibit similar behavior. If I remeber correctly, the original 5D was one of them.

Andrey, as far as ISO 100, are you finding it to be more than just ISO 200 with +1 ev comp? Your posts seem to be implying something different, which is news to me.

 

[fotografz]

Thanks Marc you know me i look at all the curve balls being thrown. Like all Pro's are jobs are to solve problems and create workarounds. Every system and every setup has something that you need to work and just good to know what some of these things are. I been working around low sync speeds all my career and just have to work at it some.

Guy, when I know I have a big group to shoot or something like that I've just tossed the Metz Potato masher in a Boda bag and strap it to the roller ... I picked up a Sony module for it ... then I'm carrying my own bag of sunlight with me

BTW, I picked up the Metz from a forum member for less than my smaller Sony flash cost me.

 

[Greg Seitz]

There are many problems one on top of another with this test. Most obvious to me f.e. - RD is not doing good job with pushing shadows and makes noise worse than it actually is and those Raw files are compressed ARWs (lossy),
which is not harmless if you want to push. In any case all this is not that relevant. Higher ISOs are more noisy just because sensor gets less and less light from camera light meter while ISO grows, so more noise. What's important here is that we have case when picture has more details and more noise and ISO 320 resolves maximum amount of details and still remains rather clean.

Of course higher ISO is noisier, that's my point, it kills shadow detail thus reducing the dynamic range.

I'm surprised you see more detail with 320 as that is clearly not the case here - look closely at the details in the bark of the tree, there are details in the 200 shot that have vanished in the 400 shot. Some of the noise is probably from sharpening with RD but the same amount was used on both files.

It would be great to have you show some examples of 320 vs 200 where the 320 files look better overall since I don't see it with these 400 vs 200 files - providing raw files would be great.

 

[peterv]

I'm glad I started this thread, there seems to be a lot of interest for this topic. I'm very happy with all the dicussion and feedback, thanks everyone.

Andrey, I think it would be very helpful indeed if you could produce some examples of your workflow in RPP with the a900 files. And also a straightforward list of the camera settings you'd advise would be much appreciated. Now I know Douglas has kindly provided us with a lot of exposure and UniWB advise, but it would be nice to have it all in one thread and coming straight from you.

Meanwhile I developed one of the testshots I did in RPP. I really think RPP deserves it's own thread on this forum. First of all, it's amazing how much detail can be pulled out of the a900 files with RPP! But these files look noisier. Now Andrey and Douglas and Mr. Borg will say that's because you didn't use the 'right' settings and underexposed. I'd like to find that out for myself, so please post examples and your settings.

Anyway, back to 100 vs 320:

Again there's more detail in the 320. But the 320 shots are certainly a lot more noisy.
320 on the left, 100 on the right.

Screenshot to show detail, screenshot to show noise difference and full shot. BTW, all my shots were f3.2 with the CZ 85.

 

[Greg Seitz]

Peter,

Have you had a chance to compare 200 vs 320?

Thanks,

Greg

 

[Quentin_Bargate]

ISO 100 is useful where you need slower shutter speeds or a reasonable aperture. In fact I wish Sony did ISO50. Of course one can use ND filters but camera makers generally seem more intersted in blathering on about their cameras high ISO capability, which for many is not a priority.

Its good to see ISO 320 produces decent results, but at the cost of more noise and reduced dynamic range. I'd only use it if I needed a faster shutter speed. I take most shots on ISO 100 or 200.

Quentin

 

[hardloaf]

Andrey, as far as ISO 100, are you finding it to be more than just ISO 200 with +1 ev comp? Your posts seem to be implying something different, which is news to me.

Ok, probably I have to explain why I don't like those below-200s. I checked Raw data for ISO100 and 200 and to my surprise found that actually there is only 1/3 EV of a difference between them in midpoint. This means that when camera set to 100 gain is actually set to 160. Camera meter though exposes it like 100. Now guys enable your long forgotten film reflexes and answer what's going to happen if we expose ISO 160 film as 100?

I find this all way too weird and complex to use when shooting. They essentially overexpose 160 for 2/3 EV and plus those clipped shadows in R and B channels make everything highly unpredictable. Thanks Sony, but no.

 

[peterv]

Hi Greg, not yet, but would like to do that in the coming week.

 

[hardloaf]

Andrey, I think it would be very helpful indeed if you could produce some examples of your workflow in RPP with the a900 files. And also a straightforward list of the camera settings you'd advise would be much appreciated. Now I know Douglas has kindly provided us with a lot of exposure and UniWB advise, but it would be nice to have it all in one thread and coming straight from you.

I'll do this in a separate thread - this one is already too long.

Meanwhile I developed one of the testshots I did in RPP. I really think RPP deserves it's own thread on this forum. First of all, it's amazing how much detail can be pulled out of the a900 files with RPP! But these files look noisier. Now Andrey and Douglas and Mr. Borg will say that's because you didn't use the 'right' settings and underexposed. I'd like to find that out for myself, so please post examples and your settings.

I'd say if you see objectionable noise in RPP which shows in final prints or down-sampled web images you need to start with decreasing sharpness slider in RPP and do selective sharpening in Photoshop or something like this. RPP doesn't do any noise filtering, so whatever you see is pretty much what your camera produces. This all is intentional and indeed is a topic for another talk.

Anyway, back to 100 vs 320:

Again there's more detail in the 320. But the 320 shots are certainly a lot more noisy.
320 on the left, 100 on the right.

Screenshot to show detail, screenshot to show noise difference and full shot. BTW, all my shots were f3.2 with the CZ 85.

I wouldn't call it 'a lot more noisy' for most of practical uses. I'll try to produce some samples, but let's first settle with few things.

1. Properly exposed shot on ISO 320 will have more noise than on ISO 200. No doubts in that. ISO 200 should also produce higher DR, if not that R/B channels shadow clipping. So it actually doesn't - all that advantage is eaten by clipping and non-linearity in R/B channels.

2. 320 is good because that's first ISO which doesn't clip shadows. Please stop repeating that you don't see it in all your hundreds of ISO200 shots - just shoot something with colored shadows at 200 and 320, process identically and compare ignoring noise. If you don't see the problem then you have nothing to worry about - your perception is in balance with Sony and go get some life

 

[hardloaf]

Here is first example.
Same exposure (shutter and aperture), different ISOs.
200 left, ISO 320 right.
Shots were underexposed, so 320 was compensated for 3 stops and 200 for 3 2/3 in converter. Also only WB and tone curve were applied without any kinds of filtering, sharpening and so on. In theory 200 should be cleaner, but actually 320 shows disproportionally better result with exactly the same amount of light reaching the sensor. 200 has purple shadows and on 100% view you can see that it's really troubled.

View attachment 16969

View attachment 16970

 

[fotografz]

ISO 100 is useful where you need slower shutter speeds or a reasonable aperture. In fact I wish Sony did ISO50. Of course one can use ND filters but camera makers generally seem more intersted in blathering on about their cameras high ISO capability, which for many is not a priority.

Its good to see ISO 320 produces decent results, but at the cost of more noise and reduced dynamic range. I'd only use it if I needed a faster shutter speed. I take most shots on ISO 100 or 200.

Quentin

I actually agree, and even wish there was ISO 25 available for some situations. I don't see this camera as a high ISO solution, so I don't use it that way (800 is about all I really need for a majority of my work, with 1000 as a reserve when needed.) Conversely, I'm not a fan of the lower ISOs on the Nikon D3/D700 ... but from about ISO 500 and above I love it.

This is why I'd like to see Sony bring out a D700 competitor ... although it's been explained to me why that probably won't happen.

I also think the noise issue at 320 is exaggerated. In 100% computer screen blow-ups it is far more apparent than in a print, even a large print. What is apparent in the ISO 320 print is the gain in micro detail which adds to the over-all dimensional feel a bit more.

Personally, I don't want the camera to lessen the micro contrast qualities that some of the Zeiss lenses bring to the party ... at the very least I want the camera to stay out of the way of those charcteristics.

 

[kuau]

Here is first example.
Same exposure (shutter and aperture), different ISOs.
200 left, ISO 320 right.
Shots were underexposed, so 320 was compensated for 3 stops and 200 for 3 2/3 in converter. Also only WB and tone curve were applied without any kinds of filtering, sharpening and so on. In theory 200 should be cleaner, but actually 320 shows disproportionally better result with exactly the same amount of light reaching the sensor. 200 has purple shadows and on 100% view you can see that it's really troubled.

View attachment 16969

View attachment 16970

So based on this test, is 320 superior even if the test was done properly exposed?

Steven

 

[hardloaf]

So based on this test, is 320 superior even if the test was done properly exposed?

Steven

That's subjective - you'll have to decide for yourself depending on what and how you shoot. There is less noise with 200, but shadows will be clipped. More noise and better shadows with 320. If shadows will need to be pushed with tone curve or exposure correction I'd definitely stay with 320.
Personally I always stay at 320 because I don't even consider its noise level to be something to worry about. I tested 320 up to 30 sec. without any in-camera NR and it's definitely non-issue for me, but it's much safer for future post-processing.

 

[jmvdigital]

Just curious what you guys think of Michael's weigh in of the issue on LL:

The camera's base ISO is 200, with ISO 100 being an extended speed. On most cameras the use of lower than native ISO is usually not a worthwhile exercise since it simply crushes dynamic range and contrast. But that's not the case with the A900. What the use of ISO 100 appears to do is reduce headroom at the top end of the exposure scale, but opens it up about a half stop in the shadows. It also lowers the noise (which is already very low at 200) from a Noise Ninja reading of 11 down to 9. According to DxOMark the noise reduction is almost 1db, from a SNR of 35.2 at ISO 200 to 36.3 at ISO 100, corroborating what is seen in Noise Ninja.

Based on an extensive series of tests, I am now shooting with the A900 at ISO 100 when I can, which means when working on a tripod and when shadows are more important than highlights.

BUT: But, be aware that by reducing the ISO below the camera's native 200 you are sacrificing one of the A900's great image quality assets, and that's its very broad highlight "shoulder". To my eye this gives the camera a very distinctive advantage in that highlights seem more film-like, with less of a tendency to clip, and to show more nuanced tonal and colour separation in lighter tones. For many current and prospective users, especially those with a sophisticated appreciation for the nuances of image quality, this highlight latitude may provide the camera's biggest appeal, so only switch to ISO 100 in situations where slightly lower noise and more shadow range are what's desired.

http://luminous-landscape.com/reviews/cameras/a900-one-month.shtml

 

[hardloaf]

Hello Justin.

Just curious what you guys think of Michael's weigh in of the issue on LL:



http://luminous-landscape.com/reviews/cameras/a900-one-month.shtml

Nothing have changed for me and I still don't use any ISO below 320. MR did the right thing though - he tested everything with his workflow and made his conclusions. I always encourage everybody to do the same - you may love one converter or another, but most of people don't realize that their converters and the whole workflow may severely limit their cameras and lenses abilities and you must adapt your shooting style, reflexes and habits to squeeze more out of your equipment. This is certainly the case for pretty much all converters, noise filters, lens corrections and so on.
RPP is designed to minimize those limitations and you can actually get pretty much everything out of your camera and lens, but there are consequences. Everybody who tried RPP knows what I mean

 

[douglasf13]

I gotta say, Andrey, I finally moved back to an all Mac computer arrangement, and I started using RPP. Wow, what a revelation. Sure, LR3 beta has come along way over LR2, and the ease of workflow is nice, but your converter is so good that I'm willing to add the extra step in there...regardless of a bit of waiting time. I'm sending my RPP made tiffs to PS/LR, and that seems like a wonderful compromise. The film sims are great, btw. Donated earlier this week!

 

[Kirk Candlish]

Okay, this made me curious. So I just ran a controlled test in studio.

Available but constant light (Profoto modeling lights in soft boxes, no variable)

You're assuming that the modeling lights are voltage regulated and therefore constant ?

Or you metered for each shot ?

I've been shooting with Profoto for over a decade and I have seen the modeling lights dim with line voltage fluctuations.

 

[hardloaf]

What can I say - welcome back to the Mac world and I'm glad that you find RPP useful

I gotta say, Andrey, I finally moved back to an all Mac computer arrangement, and I started using RPP. Wow, what a revelation. Sure, LR3 beta has come along way over LR2, and the ease of workflow is nice, but your converter is so good that I'm willing to add the extra step in there...regardless of a bit of waiting time. I'm sending my RPP made tiffs to PS/LR, and that seems like a wonderful compromise. The film sims are great, btw. Donated earlier this week!

Andrey