S3 soft? I don't think so!

[Lee Jay]

Lee...(and others)

Your contributions to this site are great, thanks! I'm patiently
(trying to be anyway) waiting for my S3 to come myself and eagerly
read all you've had to say about it, as well as Simon. I was
wondering if you know of any good books or articles online that can
be read to learn about processing photos (both camera settings and
in apps like Photoshop). I'm looking to one day of course get a
dSLR, I checked out the Olympus E-500 and the Canon XT the other
day,very nice

There are a ton of books on it. The only one I have is this one:

http://hiddenelements.com/

Also, do you have any expeirence getting Panorama's developed? I
have quiet a few I've taken over the last 6 years, but the sizes
are so odd after putting the photos together with Autopano that the
sizes available to print the photos in on the development sites are
larger than the dimensions of the photos (something will get cut
off, stretched or a white border).

I usually print them something like 8x12 or 12x18 with 2 or 3 panos on a single print and then trim them.

--
Lee Jay
(see profile for equipment)

 

[Lee Jay]

I don't see any noise on the skin or eye? What is it that you see
as "noise"?

May be it depends on the monitor and its calibration. I use a
1600x1200 LCD.

I'm looking at the original on a 1920x1200 LCD and don't see it.

With an 'Actual pixel' view, on the skin, I see small luminance
grain about 1 mm wide.

Which image are you looking at? There is no noise in the reduced shot so you should only be looking at the 100% crop of his eye.

A baby's skin would be smoother, I believe
(I don't usually look at babies from so close though). In the white
of the eye, I see very small blue spots. This might be natural, it
is the same color as the iris.

I don't see it.

More technically, using Noiseware Porfessional I get a measure of
the noise. It gives 18% high frequency noise. This is surprisingly
high; usually I get less at ISO 80. Applying the Portait noise
reduction profile gives a better picture.

On the eye or on the reduced image?

--
Lee Jay
(see profile for equipment)

 

[Frank43]

More technically, using Noiseware Porfessional I get a measure of
the noise. It gives 18% high frequency noise. This is surprisingly
high; usually I get less at ISO 80. Applying the Portait noise
reduction profile gives a better picture.

On the eye or on the reduced image?

The 18% reading was from the reduced image. For the non-reduced eye crop, Noiseware reports between 6 and 9% high frequency noise. Applying the noise reduction shows a clear change.

I presume the horizontal axis of the Noiseware measure follows the luminance. On the reduced images, the noise starts at about 18% in the darker areas and stays at 16% half way and then it goes down to 0% in the srongly lit areas.

Cropping the reduced image to cover the same areas as your crop, shows about 6% noise. So it seems that the size reduction has decreased the noise slightly. I was epxecting the opposite due to the sharpening of size reduction.

Noiseware reports its measurment with 3 lines, 2 blacks and one red. I presume one of the black is luminance and the red might be color. I wonder what is the third one.

Frank

 

[Lee Jay]

More technically, using Noiseware Porfessional I get a measure of
the noise. It gives 18% high frequency noise. This is surprisingly
high; usually I get less at ISO 80. Applying the Portait noise
reduction profile gives a better picture.

On the eye or on the reduced image?

The 18% reading was from the reduced image.

Noiseware doesn't have anyway to know the difference between noise, JPEG artifacts and small low-contrast detail.

For the non-reduced eye
crop, Noiseware reports between 6 and 9% high frequency noise.
Applying the noise reduction shows a clear change.

Sure, it may get rid of his pores and peach-fuzz.

Cropping the reduced image to cover the same areas as your crop,
shows about 6% noise. So it seems that the size reduction has
decreased the noise slightly. I was epxecting the opposite due to
the sharpening of size reduction.

The size reduction was dramatic and should have pretty-much obliterated all visible noise at this ISO. But again, it doesn't know noise from artifacts from detail - it guesses based on sophisticated algorithms. To determine it for real, you have to shoot a zero-detail subject without compression.

--
Lee Jay
(see profile for equipment)

 

[Roy NN7DX]

I usually print them something like 8x12 or 12x18 with 2 or 3 panos
on a single print and then trim them.

I just had one of those moments where I wondered where my brain has been... OF COURSE that is how to do this...
Thanks...
--
OK, I admit it... I capture souls with my camera!
Roy NN7DX

 

[Frank43]

Noiseware doesn't have anyway to know the difference between noise, JPEG artifacts and small low-contrast detail.

....

Sure, it may get rid of his pores and peach-fuzz.

Yes, may be what I see as noise is largely natural skin pores. My experience with baby's skin is rather limited. I thought it was smoother, though. Next time I see a real baby, I will look closely

Frank

 

[Lee Jay]

Yes, may be what I see as noise is largely natural skin pores. My
experience with baby's skin is rather limited. I thought it was
smoother, though. Next time I see a real baby, I will look closely

This is a very tight shot showing much more than you can see with your eyes from this distance. Plus, he's 2 years and 3 months old so his skin isn't quite as smooth as it was when he was an infant (which is so darned smooth, you can't believe it).

--
Lee Jay
(see profile for equipment)

 

[Frank43]

He is very cute as he is, pores and all...

 

[Montana500]

Lee, you forgot to inform people that you post processed with a quality sharpening/downsizing program which will render any soft photo sharp.

--
2005 Canon S2 photo gallery:
http://www.wilderness-sportsman.com/hits/hits.html

 

[Montana500]

I thought that setting contrast -2 and sharpness -2 was to
facilitate post processing, but you seem to think that even without
post processing the camera is better off with these settings -
right?

The image was heavily sharpened in a third party sharpening plug in.

--
2005 Canon S2 photo gallery:
http://www.wilderness-sportsman.com/hits/hits.html

 

[pentel]

With such a super zoom lens, I believe it's sharpest at telephoto end and soft at widest. I think you can try it out and see

 

[bluenoser23]

Lee, you forgot to inform people that you post processed with a
quality sharpening/downsizing program which will render any soft
photo sharp.

Yes he did...in his first post:

Full frame reduced with bicubic sharper:

------------

At least I think that's what you're referring to Montana.

James

 

[mamallama]

Lee, you forgot to inform people that you post processed with a
quality sharpening/downsizing program which will render any soft
photo sharp.

Yes he did...in his first post:

Full frame reduced with bicubic sharper:

------------

At least I think that's what you're referring to Montana.

It's that "no processing" that's misleading if not absolutely false.

--mamallama

 

[Lee Jay]

It's that "no processing" that's misleading if not absolutely false.

No processing but what was mentioned for the full image, and the crop has no processing at all except for the crop which is why I posted it.

--
Lee Jay
(see profile for equipment)

 

[Lee Jay]

Lee, you forgot to inform people that you post processed with a
quality sharpening/downsizing program which will render any soft
photo sharp.

I did mention it (just used Photoshop Elements Bicubic sharper) for the full size and I did no processing at all for the 100% crop (other than the crop, of course).

--
Lee Jay
(see profile for equipment)

 

[Lee Jay]

With such a super zoom lens, I believe it's sharpest at telephoto
end and soft at widest. I think you can try it out and see

I've tried it at the long end and compared to one of the sharpest lenses around and it's sharp:

http://forums.dpreview.com/forums/read.asp?forum=1010&message=18506200

I haven't done a test at the wide end yet.

--
Lee Jay
(see profile for equipment)

 

[bluenoser23]

Lee,

I just came across the most recent user review here at dpreview on the S3. This guy is a current G3 user and was not impressed with the S3 at all.

For example:
by phusebox, 31 May 06

"Low-light performance is really really bad, even with the unusable ISO800 setting. It When shooting in ISO100 for example inside my canon G3 made a good photo at 1/60s whereas the S3 needed a 1/15s shutter speed. It is 4 times as slow in low light one might say."

Here is the link to his review:

http://www.dpreview.com/reviews/read_opinion_text.asp?prodkey=canon_s3is&opinion=31898

It's amazing how people's experiences seem so polarized about the S3. I've read how it's takes great shots, etc. etc. and how it isn't worth a bag of hockey pucks. This guy compares it to my A95 and says the A95 "outclasses the S3 by a great deal". What's a fellow to believe? There must be some objective, unbiased FACTS about the quality of shots taken and performance by the S3 as compared to other Canon, non-DSLR cameras (i.e A95, A620, G6, etc.).

I know it is of course a bit unfair of me to ask, but would you be kind enough to address the comments by phusebox? Many thanks.

James

--
Bluenoser
A95

 

[Lee Jay]

Lee,

I just came across the most recent user review here at dpreview on
the S3. This guy is a current G3 user and was not impressed with
the S3 at all.

For example:
by phusebox, 31 May 06

"Low-light performance is really really bad, even with the unusable
ISO800 setting. It When shooting in ISO100 for example inside my
canon G3 made a good photo at 1/60s whereas the S3 needed a 1/15s
shutter speed. It is 4 times as slow in low light one might say."

The G3 is f2 at the wide end and ISO 100 on the G3 is something close to ISO 160 on the S3. So the G3 should be 1 2/3 stops faster at the same ISO setting than the S3 at the wide end.

I think the G6 is about the best low-light Canon P&S made. It has a fast lens, a great sensor, RAW mode and a hot-shoe. It does lack IS. The S3 has a slower lens (1 stop at the wide end, 1/2 stop at the long end), a smaller, not-quite-as-good sensor, no RAW mode and no hotshoe. But you do get IS and all that focal-length range.

The S3 is capable of good performance in low-light. I've posted about that. But it's not as good as the G-series with those fast lenses and it's no-where near as good as my SLRs, especially when I use fast primes.

I've got a nice comparison from a wedding I shot last Saturday when my wife took a shot with the S3 at ISO 800 and f2.7 (wide open) at just about the same time I took a shot with my 5D with 35/1.4L at ISO 1600 and f1.6. Let me tell you, there's no contest. Nor should there be.

People need to learn how to use their cameras within their limitations and at their sweet-spots of performance. The S3 has a very wide sweet spot covering the entire aperture range on the lens, the entire zoom range (well, I haven't really tested the wide end yet), and up to ISO 400 if properly exposed. IS really does work up to 3-stops past 1/f (35mm equivalent) and beyond. Where it is weak is at ISO 800, especially when underexposed, when you need external flash, and when a lens faster than the one on the S3 is needed.

If anyone wants to put-down the S3 for it's limitations, I'll put my 5D and 35/1.4L up against their A95, G6 or any other P&S and also claim it "outclasses the [insert camera name here] by a great deal".

The S3 does what it does pretty well. It doesn't do well what the G6/20D/5D do well.

--
Lee Jay
(see profile for equipment)

 

[Hoddy4]

LJ, Is the wedding photo taken by the S3 at 800 acceptable for small prints after going through a noise reduction program?

 

[Lee Jay]

LJ, Is the wedding photo taken by the S3 at 800 acceptable for
small prints after going through a noise reduction program?

Yes - for 4x6s. A well-exposed ISO 800 shot in broad-spectrum lighting can be good enough for an 8x10. But this was in horrible dim tungsten lighting and the exposure was close, but not perfect.

The 5D shot is probably suitable for a 20x30.

--
Lee Jay
(see profile for equipment)