Sigma Four Thirds 20D-Killer rumor - Page 7

Archived from groups: rec.photo.digital (More info?)

Frederick <nomailplease@nomail.com> writes:
> No - it was a full broadsheet edition with small print.
> Absolutely readable from corner to corner. The macro lens, more than
> sensor resolution would be the more critical issue to readability
> across the frame.

That's interesting. I tried my phone book shot with a 50/1.8 Nikkor
which is not a slouch, stopped down to f/8 or so. But conditions were
not all that well controlled. I shot handheld with the built-in
flash. Main problem was low contrast especially at the corners. Some
post processing might help with that.

I do notice there's a big difference between 300 dpi scans and 600 dpi
scans of newspaper text, especially small print like classified ads.

Thanks.

Archived from groups: rec.photo.digital (More info?)

On Fri, 27 May 2005 17:11:25 +1200, Frederick
<nomailplease@nomail.com> wrote:


>I *know* that what I get from a D70 image printed on
>my Epson R1800 absolutely *trounces* anything I
>ever got printed from 35mm ISO 50 slide film at 10x8 size.


If you're comparing a D70 image on a
modern injet printer, to an optical
print from a 35 mm slide, I agree. But
compared to well-scanned film (at 4000
or 5000 dpi) the DSLR doesn't look all
that marvelous any more. I've got
lots of film scans and DSLR captures
to back that up, at the link below.

The DSLR captures are cleaner (ie,
no grain or noise) but film holds
its own in terms of resolution.


rafe b.
http://www.terrapinphoto.com/jmdavis

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

Alan Browne wrote:

> When it comes to seeing the limits,
> measurements are what count.

If people are measuring the right data.. I haven't seen anyone comparing the
measurement of saturation limits and color separation at high saturation
levels etc. Only noise graphs..

--

Stacey

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

In article <3fn4kuF8ocaaU1@individual.net>, Stacey <fotocord@yahoo.com> wrote:

>The problem is that digital camera brands never use the same "capture
>technology" so this doesn't work exactly the same way as it did with film.
>And you surely must agree at some point enough is enough for a given print
>size or else everyone would be shooting the same size film as they wanted
>to print it? Sure for huge prints 4/3 might be at a disadvantage at some
>point but for 11X14 and under I can't imagine it ever being an issue.

I already find that my DSLR prints at that size lack sharpness if they have
foliage in, so 6 megapixels is inadequate for such medium sized prints. A4
is realistically as large as my 10D will go if the picture is to survive
close scrutiny of its fine details. YMMV.

Thing is, there are plenty of cheaper cameras that produce results which are
good enough, and which are smaller and more convinient than any DSLR. A
camera with the lack of bulk of a zoom compact and the image quality akin to
a DSLR would be a really great product to have. A camera with the bulk of a
DSLR and image quality that's getting dangerously closer to zoom-compact is
the *wrong* combination.

>> 4/3 has the smallest capture area in its market segment (digital SLRs),
>> and so will struggle to match its competitors for image quality.
>>
>
>Then again it does has some advantages being smaller like more DOF at a
>given f stop, longer lenses are smaller and cheaper for the same FOV etc.

Only the reality appears to be that Olympus' lenses are actually very
expensive, and not much lighter than the same field of view on an APS-C
DSLR.

This was supposed to be the comelling advantage of 4/3, and it's barely
there, if it's there at all. The other guys can do "small and light" as
well, using their existing lens mounts.

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

Stacey <fotocord@yahoo.com> wrote in news:3fn42jF8fq27U4@individual.net:

> Here's a great example of E300 image quality vs a 10D.
>
> http://forums.dpreview.com/forums/ [...] e=13639123

In your language does "great" mean the same as "very poor" means in
English?

I see 2 images and can't think of any way to compare them.

Maybe if the subject was the same, and the lighting, and the camera
position - then we still have 2 different lenses as well as 2 different
cameras. But it would give us a reasonable comparison.

Your example is a joke!


--
Mark Heyes (New Zealand)
See my pics at www.gigatech.co.nz (last updated 3-May-05)
"There are 10 types of people, those that
understand binary and those that don't"

Archived from groups: rec.photo.digital (More info?)

Paul Rubin wrote:
> Frederick <nomailplease@nomail.com> writes:
>
>>No - it was a full broadsheet edition with small print.
>>Absolutely readable from corner to corner. The macro lens, more than
>>sensor resolution would be the more critical issue to readability
>>across the frame.
>
>
> That's interesting. I tried my phone book shot with a 50/1.8 Nikkor
> which is not a slouch, stopped down to f/8 or so. But conditions were
> not all that well controlled. I shot handheld with the built-in
> flash. Main problem was low contrast especially at the corners. Some
> post processing might help with that.
>
> I do notice there's a big difference between 300 dpi scans and 600 dpi
> scans of newspaper text, especially small print like classified ads.
>
> Thanks.


I would have thought the 50mm 1.8 would be okay. Some D70 and other
DSLRs have had back-focus issues - I checked mine with the 105mm at F2.8
shooting a ruler on an angle at about 35cm - and it appears to be spot
on. The 105 Sigma appears to lose a little contrast at the edge, but is
uniformly sharp across the frame - as you would expect, especially with
the crop factor.
I used the built-in flash too.
Telephone book text (where I live) is a much smaller font than
newspaper. After a little USM I can read the text in the phone book
pretty well, shot at 5 metres - the same distance needed to get two
broadsheet newpapers in the frame. The letters are getting a bit
pixelated though by the time the font is large enough to read on screen.
(150%)

Archived from groups: rec.photo.digital (More info?)

Frederick <nomailplease@nomail.com> writes:
> I don't dispute that. But I'm confident that the resolution that can
> be achieved with 35mm film is seldom achieved in practice by most
> people, even when shooting low ISO slide film. And for those who do
> have the resources to get the most from film, a high end DSLR is also
> more likely to be within budget.

If I just want to shoot a few 35mm frames at ultra high resolution, I
can do that with a relatively cheap camera (my old F3 and MF 55/2.8
are practically worthless now,), a solid tripod or stand, and slow
film (I guess not Agfa Copex any more, since Agfa just announced
bankruptcy). I'm interested in DSLR's because I want to shoot in
volumes where the cost of film and scanning would kill me.

Archived from groups: rec.photo.digital (More info?)

In article <7xpsvduhyv.fsf@ruckus.brouhaha.com>,
Paul Rubin <http://phr.cx@NOSPAM.invalid> wrote:
>
>Finally there's a break; S(t)igma is going to make a 30mm/1.4 for DX
>cameras, so it can be smaller and lighter, as you say. But oops, it
>uses 62mm accessories, so it's bigger (the full frame 50mm's use 52mm
>accessories),

A 30mm lens on a typical SLR requires retrofocal optics. A 50mm typically
doesn't (assuming we're talking about 35mm-style systems here) - that's
probably got a lot to do with it.

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

Stacey wrote:

> Then show us the resolution and contrast in a 100% crop, shot with a equiv
> 300mm lens hand held one stop from wide open at infinity like my shot was,
> oh yea and upsample it so it would print at 300DPI on 8X10 before you cut
> the crop out! :-) Maybe then we can compare "image quality"?


First of all, in an image test, a tripod should be used. Handheld shots
introduce unknowns unless the shutter speed is quite high.

Was that lens a zoom or a prime?

For me, 3000 / 300 dpi works out to 10 inches right off the top, so no
problem at all, and upsampled for 12x8, no artifacts will show in a
print at all.

If your "300mm equivalent" is a zoom, then suggest a subject and I'll go
shoot it with my 80-200 f/2.8 at 200mm (300 eq.). You pick wide open or
2 down as you please. I would suggest church steeples or other detailed
architecture on a sunny day... lots of contrast to pick up texture.

For primes, I'll hapilly use my 20, 50 or 100 (30, 75, 150 eq.).

Cheers,
Alan.


--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

Frank ess wrote:

> Any word on the likelihood of photodo coming into the recent arena? When
> was the last new test? Have there been no significant changes in the
> lens population since then? I'd like to see what their methods would
> show about newer equipment.

I don't think so. The links to the charts have to be 'filled' to work
at all. No new lens data. No e-mail replies to get them to fix the
chart links. etc.

Too bad. I guess the site owner has simply gone on to other things,
left the site there.

Cheers,
Alan

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.

Archived from groups: rec.photo.digital (More info?)

In message <7xhdgp5m9u.fsf@ruckus.brouhaha.com>,
Paul Rubin <http://phr.cx@NOSPAM.invalid> wrote:

>Stacey <fotocord@yahoo.com> writes:
>> > Part of the problem is white balance. Even in white light, most
>> > digitals are much more sensitive to green than they are to red, such
>> > that the red channel is always exposed a stop below the green, making it
>> > noisier and more posterized.
>>
>> How do you "check" this problem on a specific camera model? Would you see it
>> in a RGB histogram in a RAW converter? What type of "target" would you
>> shoot to check this?

>Huh? Shoot a grey card in sunlight and look at the RGB histogram.

But where? RAW converters show you the converted image's histogram, not
the RAW histogram, so all the scaling is hidden from the user. You need
something like IRIS to see the RAW data.
--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS@no.komm>
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

Chris Brown wrote:
> In article <HmCle.40723$G8.18501@text.news.blueyonder.co.uk>,
> David J Taylor
> <david-taylor@blueyonder.co.not-this-bit.nor-this-part.uk> wrote:
>> Chris Brown wrote:
>> []
>>> I already find that my DSLR prints at that size lack sharpness if
>>> they have foliage in, so 6 megapixels is inadequate for such medium
>>> sized prints.
>>
>> I have to ask - that's not using JPEG for the image format by an
>> chance?
>
> CRW, converted to 16 bit PSD using ACR, sharpening done as the final
> step pre-priting. I don't do JPEG.

Thanks. I just wondered because loss of detail in foliage and grass is
something I've seen in heavily compressed JPEG, but I thought it unlikely
here...

David

Archived from groups: rec.photo.digital (More info?)

JPS@no.komm writes:
> If there is, what does that mean? In this particular case, I don't
> think that it's anything like running a median filter over the data, or
> running something like Noise Ninja. I believe that the so-called noise
> reduction is analog, and only counter-acts know circuit-induced
> distortion. There is also, of course, the replacement of defective
> sensels' data with interpolated data from the surrounding pixels.

Canons don't do that to their raw files. Nikons do, and it gets
astronomers ticked off because it makes the stars disappear from the
pictures.

Archived from groups: rec.photo.digital (More info?)

In message <3fnq1cF8gqdjU3@individual.net>,
Stacey <fotocord@yahoo.com> wrote:

>JPS@no.komm wrote:

>> In message <3fnaqpF8mesfU1@individual.net>,
>> Stacey <fotocord@yahoo.com> wrote:

>>>My point was that the color just doesn't look right on the canon shot. It
>>>might do a great job recording a color chart accuratly but IMHO it has a
>>>hard time dealing with highly saturated colors and blowing out reds etc.

>> Come on now, do you really think the camera has anything to do with this
>> in RAW mode?

>It has more than a little something to do with the design of the sensor.

The sensor is a panchromatic linear-capture device. The variations
between models have more to do with noise and dynamic range than
spectral response.

>> RAW data is very simple; three sets of filters, and about
>> 3900 levels for each pixel. If the camera's JPEGs or the Canon
>> converters clip saturated colors, that does not stop you from converting
>> them in such a way as to avoid the clipping. In fact, simply setting
>> the cameras contrast to -2 usually avoids that, and does not result in
>> flat, "low contrast" JPEGs.

>Yet they will look flatter than a camera that doesn't clip the colors and
>doesn't need the contrast lowered to stop it from happening.

Not especially. I started using -2 contrast recently, so that the
histogram on the back of the camera would be more accurate for RAW
shots, and the JPEGs are not really much flatter looking than with
default contrast. It is mainly the extremes that are altered; the gamma
of the midtones is only affected slightly by extreme contrast settings.

>You can't seriously think every sensor can capture colors exactly the same
>can you?

No, but the differences are probably not as big as you would make them
seem. The biggest difference, most likely, is in the CFA; not the
sensor.

Anyway, I have rarely had a flower blow-out on my Canons that didn't
disappear with a dark 16-bit conversion with curves or gamma (in levels)
applied in PP. If the RAW data isn't blown, the image is always
workable.
--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS@no.komm>
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><

Archived from groups: rec.photo.digital (More info?)

"David J Taylor" wrote:
> Chris Brown wrote:
>> David J Taylor wrote:
>>> Chris Brown wrote:
>>> []
>>>> I already find that my DSLR prints at that size lack sharpness if
>>>> they have foliage in, so 6 megapixels is inadequate for such medium
>>>> sized prints.
>>>
>>> I have to ask - that's not using JPEG for the image format by an
>>> chance?
>>
>> CRW, converted to 16 bit PSD using ACR, sharpening done as the final
>> step pre-printing. I don't do JPEG.
>
> Thanks. I just wondered because loss of detail in foliage and grass is
> something I've seen in heavily compressed JPEG, but I thought it unlikely
> here...

I suspect that there was a bit of a communication failure here. What Chris
was saying, I think, was that he finds 6MP dSLR images _in general_
inadequate to make quality 11x14s for landscapes with foliage. That's the
experience here as well. Even at A4 with the Epson R800, 4000 dpi scanned
645 shows more detail and better textures than 6MP or 8MP digital. Certainly
there are a lot of images for which 6 or 8MP will make a lovely 11x14s, but
not landscapes (or at least landscapes that depend on detail for their
impact).

David J. Littleboy
Tokyo, Japan

Archived from groups: rec.photo.digital (More info?)

"David J. Littleboy" <davidjl@gol.com> writes:
> > Do DX cameras really want to be thinner?
>
> That's why Canon does EF-S lenses<g>. Stigma has the problem that they need
> to handle the largest flange-to-film distance of any of the dSLRs.

Uh oh. This starts making me want to believe in 4/3 again :-O. :-)

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

Bill Hilton wrote:

>>Alan Browne writes ...
>>
>>The 20D is almost 40% larger than 4/3, and the D70/7D/*istD
>>are over 50% larger.
>
>
> Hi Alan,
>
> Just wondering where you're getting the sensor size numbers as I see
> various dimensions floating around ... here are the numbers I'm using,
> based on the Oly spec and a quick web search for Nikon and Canon but
> they may be wrong ... these give an even larger advantage to the 1.5x
> and 1.6x sensors ...

I get Oly 4/3 [dpreview] : 18 x 13.5 = 243

Canon 20D: [dpreview] : 22.5 x 15 = 337.5

39% more.

> 4/3 ... 17.3x13 => 225 sq-mm (source is
> http://www.olympus-esystem.com/dea/products/e1/spec/ ... some of the
> 4/3 advocates use the larger 18x13.5 but this is what Oly sez ...)

Yeah, could be. It's easier for me to look for data on the dpreview
site as the layout is common, there's the 'side by side' looks, and the
site is fast.

> 20D => 15.0 x 22.5 mm => 337.5 sq-mm so 1.5x over 4/3
>
> Nikon D70 => 23.7 x 15.5 => 367 sq-mm so 1.63x over 4/3 and 1.08 over
> 20D

I believe ya! That puts the 4/3 in an even deeper S/N hole...

Cheers,
Alan.

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

JPS@no.komm wrote:

> I believe that the so-called noise
> reduction is analog, and only counter-acts know circuit-induced
> distortion. There is also, of course, the replacement of defective
> sensels' data with interpolated data from the surrounding pixels.

Sounds like good techology..

>
>>Not a "flame" but isn't that the case? I also have seen
>>results of the in camera jpegs and various RAW converters and seen there
>>is some pretty serious NR being done to the jpegs, especially at high
>>ISO's.
>
> I don't know or care about JPEGs.

Neither do I, just wondering if this was true or not from people using
these.

--

Stacey

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

Chris Brown wrote:

> In article <3fn4kuF8ocaaU1@individual.net>, Stacey <fotocord@yahoo.com>
> wrote:
>
>>The problem is that digital camera brands never use the same "capture
>>technology" so this doesn't work exactly the same way as it did with film.
>>And you surely must agree at some point enough is enough for a given print
>>size or else everyone would be shooting the same size film as they wanted
>>to print it? Sure for huge prints 4/3 might be at a disadvantage at some
>>point but for 11X14 and under I can't imagine it ever being an issue.
>
> I already find that my DSLR prints at that size lack sharpness if they
> have foliage in, so 6 megapixels is inadequate for such medium sized
> prints.

Who was talking about 6MP?
--

Stacey

Archived from groups: rec.photo.digital (More info?)

JPS@no.komm wrote:


>
>>What type of "target" would you
>>shoot to check this?
>
> You look at the RAW data for an exposure of a grey/white surface using
> the lighting in question.

How do you do that?

>
>>And if this is the case with a certain camera, wouldn't it be a good idea
>>to filter the camera lens to a natural/neutral WB in daylight if that's
>>where you shoot most of the time?
>
> That would be ideal, if the highlights are white.
>

So how would you determine the filter to use?

--

Stacey

Archived from groups: rec.photo.digital,rec.photo.digital.slr-systems (More info?)

Lourens Smak wrote:

> In article <d74ni5$1kn$1@inews.gazeta.pl>,
> Alan Browne <alan.browne@freelunchVideotron.ca> wrote:
>
>> Lourens Smak wrote:
>>
>> > In article <d71ud4$gfv$1@inews.gazeta.pl>,
>> > Alan Browne <alan.browne@freelunchVideotron.ca> wrote:
>> >
>> >
>> >>As
>> >>pixel densities go up, the noise will go up.
>> >
>> >
>> > Funny, so far this statement has been terribly wrong. What makes you
>> > think it will be true in the future?
>>
>> Please be specific.
>
>
> a few years later Canon has crammed 3 times the number of pixels, and
> Nikon 5 times, in the same area, with less noise. Why would this trend
> suddenly reverse?

>
Might as well forget it, the people argueing this are mainly users of
slightly larger sensor 6MP cameras that KNOW the slightly larger sensor is
the ONLY thing their camera has going for it.. The fact they all refuse to
post ANYTHING shot with them says it all for me..

--

Stacey

Archived from groups: rec.photo.digital (More info?)

Lourens Smak <smak@wanadoo.nl> writes:
> a few years later Canon has crammed 3 times the number of pixels, and
> Nikon 5 times, in the same area, with less noise. Why would this trend
> suddenly reverse?

It didn't reverse, it simply slowed down or maybe stopped. If the
limiting factor is quantum efficiency as Roger Clark says, then it can
get better with improving technology, but once it reaches 100%, that's
the end.

Archived from groups: rec.photo.digital (More info?)

JPS@no.komm writes:
> If the red component of something white/grey is half of green, and blue
> is 70% of green, then you want a filter that cuts green a stop more than
> red, and blue a half stop more than red, etc (in this case, about
> magenta).

So what is the explanation for the imbalance? Different
transmittivity of the Bayer filters? More efficient CCD photon
conversion for green light--but wouldn't it be even more efficient for
blue light, and what about CMOS? Twice as much green in the histogram
because there are twice as many green pixels in the Bayer grid?
Or what?

Archived from groups: rec.photo.digital (More info?)

JPS@no.komm wrote:

> In message <3fpr6aF949gdU2@individual.net>,
> Stacey <fotocord@yahoo.com> wrote:
>
>>JPS@no.komm wrote:
>
>>>>What type of "target" would you
>>>>shoot to check this?
>
>>> You look at the RAW data for an exposure of a grey/white surface using
>>> the lighting in question.
>
>>How do you do that?
>
> Make an uncompressed DNG,

Which is done how? or see below..


>
> Or, just load the file into IRIS as a RAW file, and from the processing
> menu, subtract the blackpoint. You get the blackpoint by taking an
> image with the same ISO on the camera, and the same exposure time, with
> the cap on the lens, and the viewfinder covered. Then, look at the bell
> curve, and it's peak value is your rough blackpoint.

Where do I view this "bell curve" to find this peak value, in iris or....
and which screen/menu item shows this curve to get this number.


> Then, when you
> load an exposed image, subtract this value in the "processing/subtract"
> menu. Then, do a "CFA conversion" from the menu

You mean "convert a CFA image" from the digital camera menu?

> (this interpolates all
> the black or missing values. Now, the color you see is what the camera
> saw, roughly..
>

So then you could save this and open in PS to see which channels are
stronger? Or how would you go about figuring out which channels needs to be
tweaked with a filter?

Thanx for the info..



--

Stacey

Archived from groups: rec.photo.digital (More info?)

rafe bustin <rafe b at speakeasy dot net> wrote:

> On Fri, 27 May 2005 20:23:10 -0400, Stacey <fotocord@yahoo.com> wrote:
>
>>David J. Littleboy wrote:
>>
>>> so the idea that Canon has problems with
>>> color rendition is blatantly ridiculous.
>>>
>>
>>
>>I never said it was a "problem" and some people might prefer the way the
>>colors are renderd by a Canon? Or are you going to claim all digicams
>>record colors and saturation the same at any level and the ONLY spec one
>>needs to look at is noise?
>
>
> What's odd is that you often choose as
> talking points things that simply defy
> objective measurement.

Sure they can be measured, but the review sites don't do it. I know
technoids love graphs and numbers, but that doesn't tell everything there
is to image quality.

Ever seen a review site mention OOF image qualities? Does that mean this
doesn't exist so we shouldn't care about that part of the images quality?

An example of color separation, I was testing the CS2 RAW converter vs the
olympus studio one on a bright red, heavily saturated flower shot. No
matter what I tried, the CS2 RAW shot lost most of the details in the reds
unless I turned down the saturation a bunch (like -20), while the studio
file didn't have this problem. when both are developed at saturation levels
that detail is still easily visible, the ACR shot look flat and bland..


Does that mean because the review sites don't talk about this kinda thing,
it doesn't exist or doesn't matter? It seems very obvious to me that the
olympus software is controlling this better than ACR is dealing with it.
Then I get called "BS" on this by people who have never used one of these
cameras, used the olympus software compared to ACR on these files etc.
Kinda like you're doing? God forbid your choice of camera isn't THE BEST on
every level! I know mine is poor at high ISO and has some issues with the
in camera ESP metering, the jpg engine uses too much NR at low ISO's etc.
It's not perfect but again God forbid anyone even dare mention that maybe
another camera has somethings about it that are better?

Almost ever test shot the review goys do is of high contrast objects, low
color saturation with no gradual color shifts in saturated colors to
record.Then they chart noise and resolution only. I guess in your mind the
rest shouldn't be anything to ever test a camera on?

>
> We all know that in the digital world
> any hue can be mapped to any other.

But you're going to claim all cameras map them the same or can even record
certain hues well at all? Or that some might record the color channels more
evenly at the sensor level than others? I know this is not the case as my
last digicam would blow anything red into a blob of color with no detail
whatsoever.

> So why talk about color in this manner?

Because it's a point that maybe canon doesn't "rule" 100% on, so of course
it can't be a valid arguement in your mind. Only high ISO noise should be
tested and is the only factor to consider..



--

Stacey