A900 Metering compared to a700 and Noise Issues ?

[j_p_p]

Hello,

This is my first post, been reading this forum for quite some time and now registered.

Anyway i posted this in Dyxum and i hope i can't link it without breaking any forum rules

http://www.dyxum.com/dforum/forum_posts.asp?TID=37737 - > a900 underexposes compared to the a700 in A Priority?

http://www.dyxum.com/dforum/forum_posts.asp?TID=37783 - > a900 VS a 700 High ISO

I brought an a900 last saturday and i am finding some things regarding Metering (Noise due to Underexposure) that i am not liking in the a900, regarding the a900 vs a700.

Other users reported the same behaviour (please read the threards, there are some examples there).
Other User reported that he always have to Overespose the a900 by +2 or +3 .

The Cameralabs review says something about this Undeexposure problems also. I Find it odd that noone noticed that given the same Aperture and Shutter Speed the a900 will produce way darker (and with more noise visible due to this) than a a700.

Please read both threads for examples and discussion.

Anyone else noticed this ? Specially asking David, because he's being testing the a900 and posting those great articles ( Do you need a A900 ? and A900 as a High Iso ...)


Regards.

[Dr. Harout]

I am sure you'll find your answer here, but meanwhile allow me to greet you and say a big welcome to this wonderful forum.

[aster]

Hello J_P_P, and welcome!

A similar question was asked in an other thread about A900 underexposing. It may not answer your every question but here it is:
David recommends checking the histogram..

http://www.photoclubalpha.com/forum/vie ... =32&t=1154

Yildiz

[j_p_p]

Yes...it's kinda similar to my 2 questions...but this is troubling me because like i said in the last Answer to the Dyxum Threads... I don't know much about Sensor Designs, so...forgive me for any mistake.

Is response to

Kiklop tha wrote :
"We should distinguish two things here;
1) metering; almost each camera may act a bit differently in certain cases and some time is needed to get used to it
2) sensor exposure under same conditions; my experiences (and i may be wrong of course) tells me that a900 will underexposure compared to a700 in same situations (using same lens, same camera settings, same light). I'm talking about RAW files of course, converted with DCRAW (linear mode)
But in second case there is one thing we need to have in minds that does influence results to some extends; with a900 we need to take into account light fall off which in some cases may result in more than 1EV light difference (depends on the lens and aperture used). This is partially explaining a900 underexposure at the same settings (if evaluating entire frame). The rest (and i get underexposure even on a cropped area) is i guess lower ISO sensitivity on a900 (remember that stated ISO equivalent ratings aren't necessary (actually almost never) for raw capture but rather for what the camera is producing in jpg mode (including gamma correction, color profile etc)"

i answerered...

"Hum...that might explain it... so you are saying is that the a900 might be less sensitive to light compared to the a700 given the same ISO.
That's not very good i think ... in FF i may need to raise the ISO more frquently than in the a700 (DOF issues of being shallower in a FF body), so you actually woorried me more... because if i have to use a 1+ EV compesation to make a similar image , and the noise is already similar between the say 1600 ISO a900 and the 1600 ISO a700 (altough one may be more sensitive than the other ...thus the a900 being darker at the "same" ISO / F / Speed ) ... that doesn't seem very good
If i understood correctly....
"

What is troubling me...and making me think about returning the camera (30 Days Trial) until Sony does something about this if it ever does....
The Noise is similar at a "pixel per pixel" between the a900 and a700 (not so noticeable in the a900 if viewed ate the same size) , but if at the same settings it produces the same Noise but it's darker (Noise is MUCH more evident if you see my shots in the thread a700 v 1900 High ISO due to the Darknes) and needs a "a 1+ EV compesation" ... if i want to make the images similar by raising the ISO 1 stop or more ,i will be getting more noise if i understood correctly.

The a900 is a lot of money (i am amateur this is just a hobbie) and i was gonna to sell the a700...but i am not liking the Metering (Dark Images with Histograms go waaaaaaaaaay into the Left if i shoot believing the Machine Metering ) and the Dark images / Lots of noise visible that i am getting.
And if the explanation might be that for the same ISO the a900 is actually Less sensitive than the a700, i am not liking it.
The a700 Meterings seemed much much more reliable, i could "believe" most of the readings it made blindly ...

[pakodominguez]

Hello,

This is my first post, been reading this forum for quite some time and now registered.

Anyway i posted this in Dyxum and i hope i can't link it without breaking any forum rules

http://www.dyxum.com/dforum/forum_posts.asp?TID=37737 - > a900 underexposes compared to the a700 in A Priority?ds.

I can't download your samples (firewall too "sensitive" at work) but for what I read at Dyxum, your test lacks some rigor, like using the same lens/focal distance (since the A700 have a 1.5 "crop factor") turn off DRO and any other automatic post-processing enhancement, even AWB...

Anyway, apparently other A900's users had experienced the same issue. I can't help you now, I don't have that camera yet, and I won't have it before... February? Inch'allah!

In other hand, you can not expect both cameras behaving the same way, even if 2 cameras form the same model won't reproduce the image the same way (ghost in the shell...)

I think Sony is already working on a Firmaware upgrade, fixing some things -they are aware the software factor is their weakness. Probably they have this issue on the ToDoList too.

Regards

[pakodominguez]

One more test you can do is using a flash on manual mode and find if there are any differences. If yes, the ISO isn't right

[j_p_p]

I had DRO Off in the ISO comparsion shots. Used IDC with the same settings and as far as i can remenber the AWB was Auto in both cameras.
I Only ever tough about doing the test of a700 vs a900 , because after taking a couple of shots with the a900 i immediatly thought "I am getting darker images than before... and the histogram is getting frequently underexposed / to the left"

"One more test you can do is using a flash on manual mode and find if there are any differences. If yes, the ISO isn't right"

Hum...yes, i thought about doing something along those lines...but i haven't got any flash for the a900
But if the ISO isn't right...i am getting already similar noise levels (can't say 100% for sure due to the darker a900 image that makes noise more visible, but i'd say so ---) for the same ISO...so if i raise it

But like you say , the other people are saying and getting the same thing as me, so ... my initial problem was that there might be something wrong with my a900, but doesn't seem so.

I dont' expect both cameras to be exactly the same ligtwise /imagewise, my concerns are with the Metering giving Darker images / Getting Darker images with the same F , Shutter Speed and ISO / Noise being more visible due to the darkness / Need to raise a stop (Iso maybe with subsquent increase in noise) or more to get properlly metered images and the FF impact on Depth of Field / thinking am i getting a camera with a metering i can't fully trust (liked the a700, pretty spot on most of the times, don't want to check the histogram each and every time) ...evertything else written in those 2 threads an what i and the others said before.

Hope this is a Firmware Issue or something like that, and not a sensor hardware fault.

[j_p_p]

Another user in Dyxum used my RAWs and after increasing the a900 exposure tried processing them via ACR and dcraw

In IDC i had to increase the a900 files by 0,7 to kinda match the histograms and Apperance to the a700s

http://foto.ixbt.com/?id=photo:265054

http://foto.ixbt.com/?id=photo:265065

http://foto.ixbt.com/?id=photo:265064

http://foto.ixbt.com/?id=photo:265063

"Ok see Your's RAW in ACR and in dcraw.
In ACR RAw from A900 is underexposed on ~0.9-0.95EV.
In dcraw with -a options the histograms of A700 and A900 near...
Hmmm... ACR is practically stadart for profesionals photographers. And also after downsizing imahes from A900 to 12Mp get the same level of noise. It's mean that A900 noise comparative with A700 but not per pixel, and in print of course up to 12Mp prints.
http://foto.ixbt.com/?id=photo:265054
(sharp 0, CNR 0, LS 0)
and
after dcraw
http://foto.ixbt.com/?id=photo:265065
"

[David Kilpatrick]

This may be ISO dependent. I will do some tests when I have time, and conditions, for controlled comparisons. I found that the A700 was typically giving 0.7-1 stop less exposure than Nikon cameras, but producing the same density. The A900 does seem to need a definite +0.7 to 1 exposure at 3200/6400, but I have not noticed some much error at 200-1600. In fact my exposures at low ISO have been better, more correct, than the A700, with the 17-35 (compared to the 11-18) and 28-75mm (compared to the CZ 16-80mm). But with the 70-300mm SSM they have been on the under-side, which I don't mind at all, since so many tele shots have no sky and need minus correction to be normal.

But - aircraft and bird shooters don't need any minus correction normally, because their subjects are against sky. They will find the results too skinny.

I think that the A900 exposure maybe favours the darker colours of my surroundings in Scotland.

David

[j_p_p]

Hum... interesting .

I have to say i tried more with ISO 1600-3200-6400 because the noise since it's underexposed seems more visible.
I first noticed that a ISO 1600 shot i took had too much noise compared to the Samples i was seeing in the Net...than i noticed that the shot was definitely underexposed by 1 EV or something like that, and that's why the noise seemed worse than it was.
Then i took a couple more shots at the high ISOs i mentioned and i began noticing this. I only have 2 FF lenses the 24-70 and the 70-300 SSM (also brought Saturday).

I am no techie so excuse me, if this question doesn't sound pretty basic....

But is it possible that the a900 sensor is less sensitive to light than the a700 ate the same ISO / are calibrated differently or something like that ?

My main question could this be fixed with a firmware release like i think it is, or can be a Hardware question ?

[Greg Beetham]

I'm just wondering if the lens shading (light cone varianble from center too field edge) is done "on chip" or later down the line internally with some pre-determined data values for each lens, and consequently a possible T/stop-f/stop variance, and as a result an impact on ISO values....that is, if there "is" an adustment to even out the light cone, (I thought David mentioned this once before).
Greg
ps Just a thought.

[David Kilpatrick]

A raw file actually has no ISO speed rating. They vary a lot in terms of relative density, so the ARW from the 900 could be darker than the 700 in absolute values, and DCraw may see it as such, because it's not calibrated to apply the correct conversion curve. It is unlikely that the A900 sensor is lower in sensitivity, or noisier, than the A700 - it uses a similar architecture and has a larger sensel (pixel well) surface area. The effect you see may be a result of A-to-D conversion values, digital gain settings, and also as already mentioned of vignetting.

To make a strict comparison crop the A900 images to APS-C size. With the 70-300mm G SSM, there is at least one stop in density increase to the outer zone of the image. It has quite strong vignetting. The 24-70mm Zeiss has sharp corner vignetting but otherwise should have an even response.

I have just run a test on a neutral target, and at ISO 6400, the A900 recorded light value 2.62 versus the A700 seeing it as 2.75 - a different easily explained by different lenses. The exposure was identical (1/250 at f6.3 on the A900, versus 1/160 at f9 on the A700 - these are exactly the same EV values) and the JPEGs were identical in density. The 900 raw file was a little darker, about what would be expected from the LV differenfe and exposure tolerances. I would say, for the record, that at this LV/EV the quality of the A900 at 6400 was better than the v4 A700 especially in colour accuracy, the A700 reporting a very different WB and producing a more magenta result from raw though the in-camera JPEG was correct. It was only quick test, not helped by the card swapping caused a problem on my Mac and actually managing to corrupt one camera's card, so I won't post any images here - I'm happy that at this setting, things are within the normal 1/3rd stop tolerances either way, but the A900 would benefit from a touch of extra exposure.

Remember - even today, the tolerances for shutter accuracy, aperture accuracy, ISO accuracy, metering accuracy are all still 1/3rd of a stop. In theory you could be unlucky and get a camera/lens combination which under or overexposed by 1.3 stops! In practice, 10% is more likely but even so with four factors to consider a cumulative error of between 0.3 and 0.5 stops either way is quite possible. It's more likely that settings are accurate and small errors cancel themselves out.

David

[01af]

The exposure was identical (1/250 at f6.3 on the A900, versus 1/160 at f9 on the A700 - these are exactly the same EV values) ...

They are similar but not equal. 1/250 s at f/6.3 is EV 13.33 and 1/160 s at f/9 is EV 13.67 ... so the latter is one third of an f-stop less exposure than the former. At ISO 6,400/39° (SV 11), these exposures correspond to the brightness values BV 2.33 and BV 2.67, respectively.

Of course, taking the usual real-world tolerances of light meters, shutter speeds, and lens diaphragms into account, a difference of just a third of an f-stop hardly means anything ...

-- Olaf

[David Kilpatrick]

Sorry, you are right - 1/3 stop out. I did a few tests and they varied from exactly the same to 1/3rd either way - as you would expect.

David

[j_p_p]

"With the 70-300mm G SSM, there is at least one stop in density increase to the outer zone of the image. It has quite strong vignetting. The 24-70mm Zeiss has sharp corner vignetting but otherwise should have an even response.
"

Hum...is the 70-300 that bad in a FF body ?

So to simplify this and if i understand correctly ...what you are saying is that the lens poor perfomance (corner perfomance - Vignetting and Light Falloff ) may actually "fool" the meter to think that the scene is actually darker than is it ?
Hum...so inb theory is it possible that using Wide Metering could be an issue... but Spot or Center can work better in FF ? Of course sometimes you want to use Wide (i do, most of the times ...i use Wide almost all the time , only using AEL look in Spot mode in more troublesome scenes) , in fact i think most people use Wide usually.