Highlight recovery

Big Ga · Mar 30, 2008

Ehrik said:
You need to realise that a D300 sensor cropped to 4/3 dimensions would
have around 7.5 Mp.

I was wondering what the ratio would give. Are you certain about this number? I was just hoping it would be a little higher. The 12 coming down to 9.5 or something.

Ehrik said:
Nikon also uses a raw conversion for the in-camera JPEGs and in their
CaptureNX software that trades chroma noise for luma noise and also
has a good noise reduction. All the others seem to lag a bit behind
Nikon on this front. Do a NEF conversion with a generic raw converter
and there will be a lot more chroma noise and shadow noise.

Hmmm... I'm not sure I agree with you there. Well.. maybe a Nikon conversion will give even LESS chroma noise, but even using ACR/LR there isn't much there as is.

Ehrik said:
I'm not convinced there is much wrong with the sensor tech of the
E-3. At least, going by the review's ISO crops, it's less than a stop
behind

Yes, but the problem is the banding! if you shoot events, often these are in absolutely terrible lighting. Honestly, often I'm at ISO6400 @ f2.8 and its not always enough to avoid motion blur. Up to ISO800 is good on the E3. ISO3200 in a dark room often isn't usable because of the banding. The Nikon sensor certainly doesn't show banding at 3200.

Big Ga · Mar 30, 2008

Ok, but have you got the same shot taken with both and where you've had to do extensive pulling/pushing of the levels?

Louis_Dobson · Mar 30, 2008

thanks.

Or at least the first part does, I wondered why there was little difference between 100 and 200.

As to the second part, I thought having discrete amps allowed for superior noise processing?

Kodak are the absolute end, I agree. If I'd realised Kodak were making the sensors (I know, it was easy to find out) I'd never have bought into Oly in the first place. I'm delighted to see the back of them.

--
http://www.flickr.com/photos/acam

Raist3d · Mar 30, 2008

Big Ga said:
Raist3d said:

This is what the point and shoot level of the SR sensor can do,

Click to expand...

And that is a very powerful example. From a finepix 700 !?! wow,
pretty impressive. Thanks for posting that.

You are welcome. I can't remember if this was my F700 (I think it was) or my F710 ( could have been). But either one or the other- they both use the same sensor. The F710 had a superior JPEG engine though.

Big Ga said:
I may very well still get the Fuji. The only thing that is really
putting me me off is a few things about using it (I can't remember
the details now, but I seem to recall things like slow RAW writing
and huge uncompressed files etc), but also, while I am sure it will
indeed give excellent DR straight out of the box (in its own JPG),

The JPEG engine of the Fuji is in my opinion, the best I have seen anywhere. Maybe the E-3's matches it "in a different way." The Fuji jpegs are scary rich.. like you can post process a lot on them and they don't all appart.

Big Ga said:
I'm not sure how it works in something like Lightroom, plus..... if
the shadow noise of say the D300/D3 is significantly better, perhaps
the same thing can be achieved if you know you're going to be
shooting in difficult conditions by simply dialing in a couple of
stops of negative exposure compensation to avoid highlight clipping
and then pull everything back in post (remember I/we tend to shoot
EVERYTHING in RAW and everything gets individually tweaked anyway).

You won't be able to match the DR of the Fuji. You can go a long way with any camera doing things differently, and I am sure the D3 has nice DR, but the Fuji will give you the most. You can afford to expose for shadows a bit and reclaim the highlights. It's just scary how much DR the SR sensor has.

The other big benefit is that in JPEGS you get a "shoulder curve" towards highlights instead of abrupt clipping. This worked at the P&S level also.

Big Ga said:
This is why I'm interested in all this stuff. I'm trying to get a
handle on how far you can push/pull things.

You can do insane things with Fuji's SR. It's really the top. The next best thing I have seen is the Sigma DP1 which is also quite insane, and i wouldn't be surprised if it matched or exceeded the D3's DR.

Big Ga said:
But now I'm even more interested in getting an S5 ! The price has
come down a lot recently. Anyone know if there is a replacement due?

Probably the end of this year (update-> probably announced at the end of this year, and coming out mid-to late next year). It will be pricey as usual. I will say this about the S5 pro though:

If you need more than 6-7 megapixels of resolution- look elsewhere. This is a 6 megapixel camera. It's true the Fuji Super CCD sampling of the projected image is better than the other sensors and depending on the subject may give you a comparable detail of 7-8 megapixels (the latesr probably like a best case) of others, but it's still mainly a 6 megapixel camera (but good 6 megapixels).

I couldn't stand the interface if you are into chimping. The zooming/panning of images is slow and is the WORST interface I have ever seen of any camera, any brand. It's ridiculous. If you do a lot chimping, beware. If you don't then it's ok.

If you ever fill the buffer (less of an issue with the S5) the camera will literally Lock you out of any operation until a full flush of the buffer. ON the S3 this was the very thing that stopped me from buying one. There was no way I was going to pay so much cash and have that kind of thing, when the Olympus e-300 (yes, back then that's what it was!) could multitask.

The S5 has the same issue but a bigger buffer, and faster memory card writing. Still something to watch for if you snap fast.

You will have to turn off the extra DR if you need a decent fps drive, otherwise it's pretty slow (and it's slow to begin with either way).

But, if you want/need DR, that's the camera. And the jpegs out of it are fantastic.

Big Ga said:
Cheers
G.

--
Raist3d (Photog. Student & Tools/Systems/Gui Games Developer)
Andreas Feininger (1906-1999) 'Photographers — idiots, of which there are
so many — say, “Oh, if only I had a Nikon or a Leica, I could make great
photographs.” That’s the dumbest thing I ever heard in my life. It’s
nothing but a matter of seeing, and thinking, and interest. That’s what
makes a good photograph.'

Louis_Dobson · Mar 30, 2008

But the D3's going to walk it, that's a given. That's why it weighs as much as a brick and costs as much as a car.

However, that Oly shot has been pushed to high heaven and it hasn't fallen apart. So a) it isn't too shabby on the DR front and b) no way are you going to have a problem with DR unless you mash the pixels or screw up the exposure. What you can't do is pull. The D3 shot has been pushed to high heaven AND the sky has been pulled as far as it will go.

Here's the same thing (more or less) with a Lee .6GND to help out (note the unblown sky).

--
http://www.flickr.com/photos/acam

Louis_Dobson · Mar 30, 2008

If you need huge ISOs, don't go E3.

However, assuming one needs the sort of quality that can't be GOT at high ISOs, then the E3 really doesn't lose very much at all to APS-C.

And also, I might add, if one IS going to go to huge ISOs, then 35mmFF wipes the floor with both formats, and when it gets cheaper is what world+dog with high ISO needs will be buying.

--
http://www.flickr.com/photos/acam

Raist3d · Mar 30, 2008

The e-3 has been found to have more DR in the shadows right? So it can certainly do it better than others there. Who cares if in the end it allows you to get the same exposure/shot. That's the point of having the DR.

Just my opinion of course. And the shot, given as evidence, shows it can be done on an E-3.

--
Raist3d (Photog. Student & Tools/Systems/Gui Games Developer)
Andreas Feininger (1906-1999) 'Photographers — idiots, of which there are
so many — say, “Oh, if only I had a Nikon or a Leica, I could make great
photographs.” That’s the dumbest thing I ever heard in my life. It’s
nothing but a matter of seeing, and thinking, and interest. That’s what
makes a good photograph.'

rovingtim · Mar 30, 2008

Ehrik said:
That ciriticism comes from cramming lots of tiny pixels on a small
CCD chip
and looking at 100%, and has a good deal to do with a NR algorithm that
happily smears detail around. Few have actually bothered to measure the
noise in the raw and those who have don't seem to think Panasonic's
sensors are exceptionally poor.

Panasonic chip + oly processing is equalling BANDING, and lots of it. The Nikon iso 6400 might not look pretty, but with noise reduction I bet it's usable for small to medium prints. Even 1600 sometimes bands on the E3. I find this an outrageous IQ lapse for a camera that is supposed to be a professional camera. Banding kills the image.

Ehrik said:
You need to realise that a D300 sensor cropped to 4/3 dimensions would
have around 7.5 Mp. If Oly ask for 10Mp on the same area, it's bound to
get noisier.

Nikon went from 10mp to 12 mp and got better noise characteristics out of it. But the main thing is they don't BAND unless you push them.

Ehrik said:
Ehrik said:

While I'm dreaming: a D300 focus modual would be nice as well. So
would a well thought out user interface.

Click to expand...

This is me being irritated with how Olympus is presenting the E3, not so much with the camera itself. Olympus is prancing about going on and on about how superb their focusing is on the E3. Quite frankly, it's not even up to the competition except for how fast it focuses on a still object in bright light. So bloody what? What pro-ish camera can't quickly in bright light?

If Oly says its got a industry leading focus system, then why isn't it too much to expect one? When I bought the E3 I was expecting industry leading focus speed in low light because that is what they were implying (-2EV, worlds fastest focus system) and what I got was one that could not even keep up with the much cheaper 40D and the same priced (cheaper in some places) D300.

For some strange reason, this pi$$ed me off.

Jonas B · Mar 30, 2008

Big Ga said:
Ehrik said:

You need to realise that a D300 sensor cropped to 4/3 dimensions would
have around 7.5 Mp.

Click to expand...

I was wondering what the ratio would give. Are you certain about this
number? I was just hoping it would be a little higher. The 12
coming down to 9.5 or something.

Maybe you are talking about different things?

I think Ehrik counts the pixels left on the D300 sensor if somebody chopped it to the same size as the 4/3 sensor. Maybe there would be 7.5 MP left there, it sounds about right.

If cropping a picture taken with a D300 to the 4/3 aspect of ratio there would be close to 11MP to play with during PP.

Well... if I did the math correctly, this is mental calculation.

regards,

--
Jonas

EDIT: Interesting thread. Thank you.

Ehrik · Mar 30, 2008

Big Ga said:
I was wondering what the ratio would give. Are you certain about this
number? I was just hoping it would be a little higher. The 12
coming down to 9.5 or something.

I used the ratio of active areas

12.3 / [(23.7 * 15.5) / (17.3 * 13.9)]

The Nikon figure was from the D50 (1.53x) but I assume it can't
differ much on the D300.

Big Ga said:
Hmmm... I'm not sure I agree with you there. Well.. maybe a Nikon
conversion will give even LESS chroma noise, but even using ACR/LR
there isn't much there as is.

Here's the D2X, Nikon's JPEG vs. a conversion without any NR at all:
http://www.pbase.com/jps_photo/image/86735542/original

Not the D300 of course, but it makes a case for Nikon NR being
advanced already then. ACR does NR too, no way around it.

People looking straight at the raw data, before conversion,
http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/D300_40D_tests/
find the D300 and 40D to be very similar. Any difference in amount
and look of noise between them will in most cases be down to the
conversion. It would be interesting if someone could do a comparable
analysis on the 4/3rds sensors.

Big Ga said:
the problem is the banding!

Yes, I know, got it on my 350D. It typically has to do with how the
camera deals with the masked pixels around the imaging area that is
used for determining the blackpoint. E.g. the K10d had a sensor with
too few masked pixels around the sensor, but also the handling by
the camera of those pixels was not the best, so it had a pretty big
banding problem when brightening shadows or shooting high ISO.
(Without it, it would have been a killer for dynamic range, apparently.)

I also understand that if the camera doesn't clip the black end of the raw
data and leaves the masked pixels in the raw file, a sufficiently smart raw
conversion can suppress the banding pretty well. Adobe, e.g. apparently
only do anything about it when they think it's needed.

So, anyway, banding is typically a shared responsibility between sensor,
camera and raw conversion.

Just my two oere
Erik from Sweden
--
CA is a chromatic abbreviation

Ehrik · Mar 30, 2008

Maybe that was not clear, but it's of course that number that matters here.

Thanks for clarifying for me.

Just my two oere
Erik from Sweden
--
CA is a chromatic abbreviation

mfbernstein · Mar 30, 2008

Ehrik said:
I used the ratio of active areas

12.3 / [(23.7 * 15.5) / (17.3 * 13.9)]

The Nikon figure was from the D50 (1.53x) but I assume it can't
differ much on the D300.

Yeah, if you want to get a 4/3 sensor size crop that's 10MP, you'd need a 15.2MP Nikon APS-C-sized sensor or a 13.8MP Canon APS-C sensor. At the current rate, you won't see something like that for another 12-18 months. Perhaps the 14.6MP Samsung sensor used by Pentax and Sony would be good enough, but I wouldn't bet on it.

Ehrik said:
Here's the D2X, Nikon's JPEG vs. a conversion without any NR at all:
http://www.pbase.com/jps_photo/image/86735542/original

Not the D300 of course, but it makes a case for Nikon NR being
advanced already then. ACR does NR too, no way around it.

The D2X was known as something of a noise-box. Doesn't really say much about the actual sensor characteristics of the D300.

Ehrik said:
People looking straight at the raw data, before conversion,
http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/D300_40D_tests/
find the D300 and 40D to be very similar.

Most comparisons I've seen show the 40D has less noise when comparable NR is applied.

Ehrik said:
So, anyway, banding is typically a shared responsibility between sensor,
camera and raw conversion.

Well, it sure would be nice if at least Oly stepped up and did something about in their conversion software. At the moment it does them no favors.

--
MFBernstein

'Wilderness is not a luxury but a necessity of the human spirit.' - Ed Abbey

Jay Turberville · Mar 30, 2008

rovingtim said:
There are two ways to create headroom. One is as you say: different
colour clipping points. But the other is where there manufacturer
decides to clip the jpg. Clip it lower on the total histogram and
there's a LOT more headroom. Canon seems to do this.

I just looked at the step wedge results for the Canon 40D and the Olympus E-330. The Canon 40D seems to show a response that is unclipped in all three channels at best about 1/3 stop sooner the the E-330. It could be less since a 1/3 stop step wedge can only show such things within a 1/3 stop level of accuracy.

rovingtim said:
Oly pushes everything as far to the right as is possible so there is
no extra information in PP. The problem is there is also more noise
in the shadows than the competition (When has a Panasonic sensor ever
been known for low noise?) so lifting the shadows creates problems,
especially at high ISO.

Well, I haven't tested the E-3, but all indications so far are that it should have the same DR as the E-330 (Panasonic claims it their newest 10Mp sensor have the same photodiode size as their 7.5Mp sensors). And the E-330 is slightly better with dynamic range as determined by noise in the shadows than the 40D. This is true down to about 10 stops from fully saturated. After ten stops, the 40D starts to have an edge. But frankly, at 10 stops, noise is pretty visibly ugly on both of them if that level is boosted out of the deep shadows.

This PNG file demonstrates that. Both images were developed with no sharpening and no noise reduction . The 40D was given a slight advantage in that its images was downrezzed to match the E-330 image. That tends to reduce noise a bit. Visibly and via measurement the E-330 has the edge down to 10 stops.

http://www.jayandwanda.com/photography/40D/DR/40D_v_E330_ShadowNoise.png

rovingtim said:
This problem is compounded by the fact that Oly has the darkest
midrange tones (at default) of any manufacturer. [Me thinks Oly is
trying to hide noise.]

Methinks you are wrong. You seem to be evaluating based on JPEGs. If so, then methinks you may be responding to other cameras using more noise reduction. But I haven't tested that, so it is only a guess.

rovingtim said:
What Olympus needs to do is pull there finger out an pony up for a
state-of-the-art sensor.

I'm pretty sure the E-3 is. I sure wish someone in the Phoenix area would contact me so that I could test an E-3 and find out. If the E-3 sensor performs as well as the E-330 sensor, then it would match or better the 40D sensor (and I suspect it does).

rovingtim said:
Sensors are expensive and perhaps they thought going small would make
them more competitive.

I think that may have been a factor.

rovingtim said:
They forgot that going atypical also means
smaller production runs. I bet Nikon much larger cropped sensor
actually cost them less than the smaller but noisy Panasonic sensor.

I doubt they forgot that. I suspect that very fact is why they wanted the lower cost factor of a somewhat smaller sensor.

rovingtim said:
Kudos to Oly for getting such great IQ out of Panasonic noise. Severe
beatings for not using the best available sensors.

Severe beatings for you for not doing more careful evaluations and for painting all Panasonic sensors with a broad brush.

rovingtim said:
Imagine an E3 with D300 Sony noise characteristics.

I don't know about D300, haven't tested one. But as for the Canon 40D, I think its there. And if not, the difference is very small.

--
Jay Turberville
http://www.jayandwanda.com

Geoff_R · Mar 30, 2008

Jay Turberville said:
I'm pretty sure the E-3 is. I sure wish someone in the Phoenix area
would contact me so that I could test an E-3 and find out. If the
E-3 sensor performs as well as the E-330 sensor, then it would match
or better the 40D sensor (and I suspect it does).

I've only done a real world testing...pics....and the e3 doesnt quite match the 40d....diff is small though.
best
--
Geoff Roughton

'Always look on the bright side life...'

Geoff_R · Mar 30, 2008

Jono Slack said:
So, I keep going all around in circles, and I get back to the E3 with
the 12-60 zoom - which works very well for me, great colour, great
ergonomics, great lens. . . . Then I think I'd like more resolution,
together with a camera which takes the Zeiss ZF lenses and off I go
again!

we've given up looking for the mythical "do-it-all" camera/system and use a variety of cameras for different tasks.....funnily enough we're shooting MF again on a fairly regular basis....we're not shooting enough yet to invest in MF digital and I'm not even sure I would if we were. Something about the tonality and lattitude of a good neg that seems lacking in digital....
best
--
Geoff Roughton

'Always look on the bright side life...'

Ehrik · Mar 30, 2008

mfbernstein said:
Ehrik said:

I used the ratio of active areas

12.3 / [(23.7 * 15.5) / (17.3 * 13.0)] [not 13.9, that was a typo]

Click to expand...

Yeah, if you want to get a 4/3 sensor size crop that's 10MP, you'd
need a 15.2MP Nikon APS-C-sized sensor ...
Perhaps the 14.6MP Samsung sensor used by
Pentax and Sony would be good enough, but I wouldn't bet on it.

Sony aren't using Samsung's sensor. Firstly, they make their own, and
secondly, the A350 has a CCD, not CMOS (and I think the exact pixel
count differs too).

The A350 has not been praised for its high ISO on the Sony SLR Talk forum.
The Samsung sensor, being MOS, seems better for high ISO
but has limited DR at ISO 100, probably due to a noisy analogue/digital
converter. Much less banding than the K10D, though.

The Samsung sensor would make a more fair 100% view
comparison to the 10Mp Matsushita NMOS as the former has only
slightly larger pixels.

mfbernstein said:
The D2X was known as something of a noise-box. Doesn't really say
much about the actual sensor characteristics of the D300.

True. I showed that to demonstrate how good Nikon's
JPEG enigine is for handling noise. And it's probably even better
now. I wish I could link to something more recent.

Just my two oere
Erik from Sweden
--
CA is a chromatic abbreviation

Jono Slack · Mar 30, 2008

HI Geoff

Geoff_R said:
Jono Slack said:

So, I keep going all around in circles, and I get back to the E3 with
the 12-60 zoom - which works very well for me, great colour, great
ergonomics, great lens. . . . Then I think I'd like more resolution,
together with a camera which takes the Zeiss ZF lenses and off I go
again!

Click to expand...

we've given up looking for the mythical "do-it-all" camera/system and
use a variety of cameras for different tasks.....funnily enough we're
shooting MF again on a fairly regular basis....we're not shooting
enough yet to invest in MF digital and I'm not even sure I would if
we were. Something about the tonality and lattitude of a good neg
that seems lacking in digital....
best

Film is out!!! I spent October 2006 shooting only film - 40 rolls, all carefully scanned. It was an interesting excercise, and while I accept that it has it's charms, the downsides were to big for me to contemplate going there again.

I'm sure you're right about the mythical 'do-it-all' camera . . . but I don't have terribly strict requirements:
1. great colour
2. great resolution
3. great lenses
4. great ergonomics
5. great weatherproofing
6. great image stabilisation
7. small size.

what's the problem?

(I don't even mind if it's expensive)

The M8 covers a lot of my needs - especially with respect to people and candids, and I like them for the rare weddings I do as well. My only criticism of 4/3 is that I would like more resolution (I'm positive that it could be done at the low ISO I'm happy with as well).

Anyway, I shouldn't complain, I just spent an afternoon with the dog and the E3/12-60, with the 50-200 and 7-14 in a small billingham bag. As this thread proves, overexposing is a bad idea, but otherwise it does a grand job.

all the best
--
Jono Slack
http://www.slack.co.uk

Raist3d · Mar 30, 2008

Geoff_R said:
rovingtim said:

Don't think I haven't thought about it. But I don't have that kind of
money to splash around. A couple of bodies I could afford, but the
lenses ... nothing wrong with Oly lenses and they have me hooked ...
for now.

By the way, the E3 focus isn't the best ...

[what is the bloody point of advertising to pros that the AF is the
fastest focus system in GOOD light ... what camera can't perform in
good light ... and then deliver a camera that isn't competitive in
CAF, isn't competitive in low light. I hate Oly marketing and
promises ... load of boll*cks, it really is.]

... but it is in the ballpark. At least I feel like I'm in the same
league.

Click to expand...

reply....we've done a load of back to back testing of the e3 V 40d
and our findings mirror most of the negative comments made here ref
the e3.
It was a very large step for us to move away from oly given the gear
we had but its been entirely vindicated.....

Don't quite know how you can say that by the reviews at dpreview, but different needs & experience for sure, may very well do so.

I think weather sealing + IS + superb glass counts for something.

Geoff_R said:
best
--
Geoff Roughton

'Always look on the bright side life...'

--
Raist3d (Photog. Student & Tools/Systems/Gui Games Developer)
Andreas Feininger (1906-1999) 'Photographers — idiots, of which there are
so many — say, “Oh, if only I had a Nikon or a Leica, I could make great
photographs.” That’s the dumbest thing I ever heard in my life. It’s
nothing but a matter of seeing, and thinking, and interest. That’s what
makes a good photograph.'

Jono Slack · Mar 30, 2008

HI Raist

Member said:
I think weather sealing + IS + superb glass counts for something.

Add excellent colour.

I quite agree, but these guys are doing events, and the want to catch the moment and then get the best out of it - even if they've seriously messed up on the exposure. Under those conditions I'm pretty sure I wouldn't be using Olympus either.

I've gone round and round in circles, and I always end up back with an E3/12-60.

all the best

--
Jono Slack
http://www.slack.co.uk

Geoff_R · Mar 30, 2008

Jono Slack said:
Geoff_R said:

Jono Slack said:

So, I keep going all around in circles, and I get back to the E3 with
the 12-60 zoom - which works very well for me, great colour, great
ergonomics, great lens. . . . Then I think I'd like more resolution,
together with a camera which takes the Zeiss ZF lenses and off I go
again!

Click to expand...

we've given up looking for the mythical "do-it-all" camera/system and
use a variety of cameras for different tasks.....funnily enough we're
shooting MF again on a fairly regular basis....we're not shooting
enough yet to invest in MF digital and I'm not even sure I would if
we were. Something about the tonality and lattitude of a good neg
that seems lacking in digital....
best

Click to expand...

Film is out!!! I spent October 2006 shooting only film - 40 rolls,
all carefully scanned. It was an interesting excercise, and while I
accept that it has it's charms, the downsides were to big for me to
contemplate going there again.

I think the big one for us was the slowing down of the process and reintroducing some "thinking" before blazing away....of course you can do that with digi but you MUST do it with film....

Jono Slack said:
I'm sure you're right about the mythical 'do-it-all' camera . . . but
I don't have terribly strict requirements:
1. great colour
2. great resolution
3. great lenses
4. great ergonomics
5. great weatherproofing
6. great image stabilisation
7. small size.

what's the problem? (I don't even mind if it's expensive)

Think ya got mate its called an.....E3

best
Geoff Roughton

'Always look on the bright side life...'

Highlight recovery

Veteran Member

Veteran Member

Veteran Member

Raist3d

Guest

Veteran Member

Veteran Member

Raist3d

Guest

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Raist3d

Guest

Veteran Member

Veteran Member

Keyboard shortcuts