DSLR picture quality limits for landscape photos?
Nils Rostedt
EOS 300/Rebel) and would like some advice from DSLR users regarding the
image quality I can expect from landscape photos.
My photographing style is on the casual side, meaning I mostly use 200-400
ISO print film handheld, sometimes 100 ISO when I feel like lugging around
with the tripod. Yes, I know I should use slow slide films to get the best
results, but this is enough for my ambitions as I also shoot other subjects.
I get decent 12x18" prints where the main quality issue is film grain in sky
areas (probably not surprising anyone). Sharpness is OK for me, my lenses
are Canon's "next level" up from the package zooms.
With the DSLR (10D), as far as I understand at least the grain/noise should
be significantly improved. But what other changes can I expect in picture
quality? Will I be able to make bigger enlargements? Other pros and cons to
think about?
Thanks in advance,
Nils
Kenwood
imagine with film.
Shooting in the DLSR's RAW mode gives you about a 2-3 stop latitude as well
unlike slide film.
Go to photosig.com and sign up.. it's free to look at pictures. Browse the
thousands of photos
that are sorted by camera.
"Nils Rostedt" <sailor...@dlc.fi> wrote in message
news:bcsqqf$ms8tc$1...@ID-88261.news.dfncis.de...
Mxsmanic
> I'm considering the step to a DSLR (in my case to
> the Canon EOS 10D from the EOS 300/Rebel) and would
> like some advice from DSLR users regarding the
> image quality I can expect from landscape photos.
Landscape photography is one of the types of photography least suited to
the use of a DSLR. If you really must have digital capture for
landscapes, use a scanning back. Otherwise, stick with film.
--
Transpose hotmail and mxsmanic in my e-mail address to reach me directly.
Mark Weaver
"Mxsmanic" <mxsm...@hotmail.com> wrote in message
news:hd74fv4l1mk04k369...@4ax.com...
> Nils Rostedt writes:
>
> > I'm considering the step to a DSLR (in my case to
> > the Canon EOS 10D from the EOS 300/Rebel) and would
> > like some advice from DSLR users regarding the
> > image quality I can expect from landscape photos.
>
> Landscape photography is one of the types of photography least suited to
> the use of a DSLR. If you really must have digital capture for
> landscapes, use a scanning back. Otherwise, stick with film.
>
Not at all -- learn to do stitched panormas with your digital images.
Instead of roughly 2000x3000, you'll have 3000x5000 or 3000x8000 or whatever
width you like. You'll be able to go 15" wide by however long you like at
200 dpi. I've had 12 x 36 - 12 x 48 pano prints made from slightly lower
res than that and the results are excellent.
Mark
Bart van der Wolf
"Mark Weaver" <wea...@nospam-corvusdev.com> wrote in message
news:_sqdnVgek4e...@comcast.com...
That would depend on the type of landscape photography. Wide scenic
overviews would be candidates for stitching. Foreground or main detail
(grass, branches), needed to add a sense of depth will be moving in the
wind. Not impossible to stitch, but hard.
Bart
Bart van der Wolf
>
> > I'm considering the step to a DSLR (in my case to
> > the Canon EOS 10D from the EOS 300/Rebel) and would
> > like some advice from DSLR users regarding the
> > image quality I can expect from landscape photos.
>
> Landscape photography is one of the types of photography least suited to
> the use of a DSLR. If you really must have digital capture for
> landscapes, use a scanning back. Otherwise, stick with film.
I agree in the sense of needing high resolution. If resolution is not
adequate for the output size, the image will look mushy. I disagree on the
use of a scanning back, e.g. because of moving branches, grass and clouds.
But on non-moving types of landscape, more reolution is better.
Bart
Bob Hatch
>
> > I'm considering the step to a DSLR (in my case to
> > the Canon EOS 10D from the EOS 300/Rebel) and would
> > like some advice from DSLR users regarding the
> > image quality I can expect from landscape photos.
>
> Landscape photography is one of the types of photography least suited to
> the use of a DSLR. If you really must have digital capture for
> landscapes, use a scanning back. Otherwise, stick with film.
>
--
http://www.bobhatch.com
Our web site about RV Stuff
A work in progress
Steve Hoffmann
example, using Canon's 16-35 L lens at 16mm, (for a 25mm equivalent view)
images with small detail towards the side of the image will look a bit
softer than film scans from an SLR with 24mm lens and fine grained films.
Ultra wide angles are not as sharp as wide angles and magnification of the
image components on the focused image plane is less with the 16mm & DSLR
then a film SLR & 24mm so there is less detail available to capture. Moving
up to 'normal' wide angle view equivalents of 35mm and greater the 'first
generation' grainless and noiseless (at ISO 100) 10D image begins to equal
or surpass fine grained film scans. See
http://www.sphoto.com/techinfo/dslrvsfilm.htm
Cheers, Steve
Surfworx Photography
shot I've seen locally enlarged to this size. I also did a sunset at this
size, the customer just collected it: he's very impressed.
Derrick
www.surfworx.com.au
"Nils Rostedt" <sailor...@dlc.fi> wrote in message
news:bcsqqf$ms8tc$1...@ID-88261.news.dfncis.de...
Mxsmanic
> Not at all -- learn to do stitched panormas with
> your digital images.
Or learn to paint in oils.
Why go to the trouble, when LF film will do the same thing in a single
exposure?
> Instead of roughly 2000x3000 ...
A good scan of LF will be more like 60,000 x 48,000 pixels.
> You'll be able to go 15" wide by however long you like at
> 200 dpi.
One use for LF landscapes is in creation of front-projection plates for
cinematography. The "print" size may be sixty _feet_ wide in such
cases, or about 2300 times larger than the size you suggest.
Mxsmanic
> I disagree on the use of a scanning back, e.g.
> because of moving branches, grass and clouds.
I suggested that only for those who cannot bear the thought of shooting
film. It's still about 16 times less resolution than 8x10 LF, though.
John Navas
In <b665fvsskbp1f3n97...@4ax.com> on Fri, 20 Jun 2003 07:20:47
+0200, Mxsmanic <mxsm...@hotmail.com> wrote:
>Mark Weaver writes:
>> Instead of roughly 2000x3000 ...
>
>A good scan of LF will be more like 60,000 x 48,000 pixels.
How silly, since the resolving power is an order of magnitude less.
--
Best regards,
John Navas <http://navasgrp.home.att.net/>
[PLEASE NOTE: Ads belong only in rec.photo.marketplace.digital,
not rec.photo discussion groups, as per <http://rpdfaq.50megs.com/>]
Nils Rostedt
">
> > Not at all -- learn to do stitched panormas with
> > your digital images.
>
> Or learn to paint in oils.
>
> Why go to the trouble, when LF film will do the same thing in a single
> exposure?
>
Thanks to everybody who answered so far. I get the impression that the image
quality of the 10D for landscapes roughly equals 100 ISO print film quality,
which seems to be sufficient for my ambitions. I still like further comments
on this topic. But please keep the topic focused on 35 mm vs DSLR - I'm
really not considering MF or LF gear at this point, even if they are the
obvious choice for specializing in landscapes.
Nils
Terje Mathisen
>>the use of a DSLR. If you really must have digital capture for
>>landscapes, use a scanning back. Otherwise, stick with film.
>>
>
> Not at all -- learn to do stitched panormas with your digital images.
> Instead of roughly 2000x3000, you'll have 3000x5000 or 3000x8000 or whatever
> width you like. You'll be able to go 15" wide by however long you like at
> 200 dpi. I've had 12 x 36 - 12 x 48 pano prints made from slightly lower
> res than that and the results are excellent.
With PanoramaTools you can do multi-row mosaics. I even use this on
fairly simple landscapes, where I want a normal 4:3 or 3:2 format. I
shoot with maximum tele instead of wide-angle, and take enough images to
cover the intended target area.
Taking 4 images in a 2x2 grid with my 4 Mpix camera gives me about 8-12
Mpix effective resolution, depending upon the amount of overlap.
Terje
--
- <Terje.M...@hda.hydro.com>
"almost all programming can be viewed as an exercise in caching"
J...@no.komm
John Navas <spamf...@navasgroup.com> wrote:
>
>In <b665fvsskbp1f3n97...@4ax.com> on Fri, 20 Jun 2003 07:20:47
>+0200, Mxsmanic <mxsm...@hotmail.com> wrote:
>
>>Mark Weaver writes:
>
>>> Instead of roughly 2000x3000 ...
>>
>>A good scan of LF will be more like 60,000 x 48,000 pixels.
>
>How silly, since the resolving power is an order of magnitude less.
I was just about to ask something along that line; my question was going
to be, "would it need an anti-aliasing filter, or would the lens take
care of that?"
--
<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <J...@no.komm>
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><
Mxsmanic
> How silly, since the resolving power is an order
> of magnitude less.
Digital is much worse than just an order of magnitude less.
--
Bart van der Wolf
<J...@no.komm> wrote in message
news:u117fvoo5fltcq22p...@4ax.com...
> In message <FvxIa.3981$%3.22...@typhoon.sonic.net>,
> John Navas <spamf...@navasgroup.com> wrote:
>
> >
> >In <b665fvsskbp1f3n97...@4ax.com> on Fri, 20 Jun 2003
07:20:47
> >+0200, Mxsmanic <mxsm...@hotmail.com> wrote:
> >
> >>Mark Weaver writes:
> >
> >>> Instead of roughly 2000x3000 ...
> >>
> >>A good scan of LF will be more like 60,000 x 48,000 pixels.
> >
> >How silly, since the resolving power is an order of magnitude less.
>
> I was just about to ask something along that line; my question was going
> to be, "would it need an anti-aliasing filter, or would the lens take
> care of that?"
Don't worry, Photoshop apparently chokes on >32K pixel dimensions.
Bart
J...@no.komm
Mxsmanic <mxsm...@hotmail.com> wrote:
>John Navas writes:
>> How silly, since the resolving power is an order
>> of magnitude less.
>Digital is much worse than just an order of magnitude less.
How much of film's resolution is just statistical coincidence; an
illusion? Just because it can resolve a certain number of lines per
inch, this doesn't mean that it resolves everything at that resolution.
It just means there are more exposed film grains where the white parts
of the lines are, showing where the lines are, but most of what exists
at that resolution is actually grain noise. Digital with its rigid
arrays is more consistent in its resolution. Once CCDs start resloving
as many lines per inch as film, they will have a lot more real,
consistent detail than film does at the same number of lines per inch,
not just enough "hints" to see the lines.
I don't accept "lines per inch" as a final measure of resolution, as
there is more than one way to obtain a line.
Charlie
wrote:
> HDigital with its rigid arrays is more consistent in its resolution.
> Once CCDs start resloving as many lines per inch as film, they
> will have a lot more real, consistent detail than film does at the
> same number of lines per inch, not just enough "hints" to see the lines.
The problem with digital is the rigid array you describe.
What you see in film as "grain" are actually grain clouds in B&W and dye
clouds in color. These are made up of dozens or hundreds of actual
molecular "grains."
These clouds are initiated exactly where needed to a definition of many
times smaller than the perceived grain. The rigid structure of the
sensors in digital is why things like fine, curved fur appears
stairstepped in digital photos.
Tou can get excellent LP/mm on vertical and horizontal lines with
digital sensors that are aligned that way. Diagonals are a horror. I did
the standard USAF test target with my 2MP Oly 2020. I forget the exact
results, but they were, I believe 15 LP/mm. I know it was in the teens,
so it can't have been more than 19. You can factor up from there for
higher MP sensors.
Film creates each grain of the size (within a given ISO) and position
needed. Straight, diagonal, tapered, doesn't matter.
--
Charlie Dilks
Newark, DE USA
John Navas
In <0s47fvshgirritjkt...@4ax.com> on Fri, 20 Jun 2003 23:17:30
GMT, J...@no.komm wrote:
True. A better method is MTF. But that's still resolution.
John Navas
In <cdilks-E10583....@news.fu-berlin.de> on Fri, 20 Jun 2003
19:52:29 -0400, Charlie <cdi...@bellatlantic.net> wrote:
>In article <0s47fvshgirritjkt...@4ax.com>, J...@no.komm
>wrote:
>
>> HDigital with its rigid arrays is more consistent in its resolution.
>> Once CCDs start resloving as many lines per inch as film, they
>> will have a lot more real, consistent detail than film does at the
>> same number of lines per inch, not just enough "hints" to see the lines.
>
>The problem with digital is the rigid array you describe.
>What you see in film as "grain" are actually grain clouds in B&W and dye
>clouds in color. These are made up of dozens or hundreds of actual
>molecular "grains."
>
>These clouds are initiated exactly where needed to a definition of many
>times smaller than the perceived grain. The rigid structure of the
>sensors in digital is why things like fine, curved fur appears
>stairstepped in digital photos.
>
>Tou can get excellent LP/mm on vertical and horizontal lines with
>digital sensors that are aligned that way. Diagonals are a horror. I did
>the standard USAF test target with my 2MP Oly 2020. I forget the exact
>results, but they were, I believe 15 LP/mm. I know it was in the teens,
>so it can't have been more than 19. You can factor up from there for
>higher MP sensors.
The standard USAF test target isn't a terribly good way to measure digicams.
That said, your Oly 2020 is also limited by its lens, not just the sensor.
Charlie
John Navas <spamf...@navasgroup.com> wrote:
> The standard USAF test target isn't a terribly good way to measure digicams.
> That said, your Oly 2020 is also limited by its lens, not just the sensor.
No. It was limited by the pixel size and stairstepping relative to line
size. It got to that point WELL before lens resolution effects showed up.
John Navas
In <cdilks-3A7A11....@news.fu-berlin.de> on Fri, 20 Jun 2003
Like many (most?) consumer digicams, the lens of the 2020 is a significant
factor in image quality.
Charlie
> >No. It was limited by the pixel size and stairstepping relative to line
> >size. It got to that point WELL before lens resolution effects showed up.
>
> Like many (most?) consumer digicams, the lens of the 2020 is a significant
> factor in image quality.
Not when you're dealing with a 2MP camera.
I've done many lens tests up to 120LP/mm with my Tamron 90MM SP where
the lines disappeared into the film grain. Also when the lens resolution
caused the line pairs to become indistinct.
The lines definitely were lost in pixels.
Mxsmanic
> How much of film's resolution is just statistical
> coincidence; an illusion?
The same amount (proportionately) that is illusory for digital images,
although for different reasons. But the absolute resolution is still
much higher, so it doesn't matter.
> Just because it can resolve a certain number of lines
> per inch, this doesn't mean that it resolves everything
> at that resolution.
Neither do digital cameras.
> It just means there are more exposed film grains
> where the white parts of the lines are, showing where
> the lines are, but most of what exists at that
> resolution is actually grain noise.
Not so. Resolution is resolution.
> Digital with its rigid arrays is more consistent in its
> resolution.
Again, not true. Results vary wildly depending on the orientation and
form of the target.
> Once CCDs start resloving as many lines per inch
> as film ...
If they ever do, although anything is possible, I suppose.
In any case, they are not there yet, so what might happen in the future
is not a relevant consideration for acquisition today.
> I don't accept "lines per inch" as a final measure
> of resolution, as there is more than one way to obtain
> a line.
You may accept or reject anything you wish, of course.
Mxsmanic
> The standard USAF test target isn't a terribly
> good way to measure digicams.
You have it backwards: Digicams don't do terribly well on a standard
USAF test target.
You don't optimize targets to favor cameras, if you want real resolution
and not just marketing blurb.
Charlie
> John Navas writes:
>
> > The standard USAF test target isn't a terribly
> > good way to measure digicams.
>
> You have it backwards: Digicams don't do terribly well on a standard
> USAF test target.
> You don't optimize targets to favor cameras, if you want real resolution
> and not just marketing blurb.
Furthermore, On the test targets used on dpReview the 2020 craps out at
around 9.5 and the 5050 craps out at around 15. I didn't read enough,
but I'm guessing that they're LP/mm since they agree with my findings
(now I believe I found around 12 LP/mm, not 15). I was being generous
due to my faulty memory. I searched my old computer and just can't find
them. Phil's nicely prove my point though.
Simple math bears out these figures.
I enjoyed my 2020 and love my 5050. I don't know why some digicam owners
can't accept them for what they are instead of bending science to make
them seem better than they are.