Minimum pixel count requirement for sharp A2

[Gregory]

After reading the Sharpness and DOF articles, I have to wonder if my camera has enough pixels to produce a nice sharp clear photo printed at maximum size on the Epson 3800 (which I've just purchased).

My Canon 350D 8-megapixel camera produces images 3456 x 2298 pixels from a 22.2 x 14.8mm sensor. Are there enough pixels for a 24x16" image? (ignoring the obvious fact that cropping the image will reduce the pixels and accentuate the problem).

Would the new 400D with 10 megapixels improve the situation significantly?

regards,
Gregory

[Henrik Paul]

This has been discussed already in this thread

[Tim Gray]

I assume you haven't taken delivery of the 3800 yet?  Otherwise you have the camera and the printer and you can decide yourself.  

But to give some response to your questions.  I print 8 mpx on an epson 4000 - at 16x24.  Without uprezzing that's about 150 ppi.  Bruce Fraser in his Image Sharpening book suggests that about 180 ppi is the threshold he applies (he doesn't like to upsample).  Regardless, I do believe that 8 mpx is pushing the limits of 16x24, but I'm not as critical as Bruce and so I accept that quality.  One day I'll have to compare 150 or 180 ppi printed natively vs uprezzed...

Increasing 8 to 10 mpx will give only a very marginal increase in quality - someone a while ago suggested that you need a 50% increase to make a  noticeable difference - and I agree.  I have a 10 mpx sony R1 and it's very close to the 1d2.  Without uprezzing that's 160 ppi for 16x24 - not much different from the 150 ppi based on a 25% increase in pixels.  

The quality of your post processing workflow will make a bigger difference than 8 vs 10.

[Gregory]

Dear Henrik,

I looked at the thread you mentioned but it seems to contain a lot of nonsensical (i.e., unrelated) mathematics and chatter. and the question asked in the inverse of my question. where I'm asking if 8 megapixels is enough to produce a sharp clear 24x18" image on my printer, the author of that thread is asking if the printer is good enough to show everything that their camera can provide.

but thank you for the reference anyway. I appreciate you're help.


Dear Tim,

I already have the 3800 and I think it's a great printer, much better than my previous Epson 1290 and it's almost certainly good enough for my needs and desires (more desires than needs ;-)

based on the theories of acutance and sharpness, I wonder if 8 megapixels contains enough information for a 24x18" image. I've seen an Ilford B&W 8x10" enlargement here in Hong Kong and the image was wonderful; sharp and clear. I would like to get a similar quality with my camera on my printer. I might be asking too much but I still want to try.

10 megapixels might only offer a marginal increase in quality over 8 megapixels but it might be worth it anyway. 15 megapixels (i.e., 50%) would apparently be *much* better but I'll have to wait until the 450D is released to get that ;-)    even the hypothetical 5D mark II to be released soon wouldn't help because the 1.6x factor of the 400D that is so useful when shooting birds would be lost and I wouldn't be getting full use of the pixels that the new 5D provides unless I could get so much closer to the birds.

thank you all. I'll probably get the 400D and then upgrade again when something with more pixels (for the same focal length/area) gets released next year...

regards,
Gregory

[Tim Gray]

most review sites will have full res sample images from the cameras reviewed - download some, print and take a look.

my favourite is the school house on Steve's...  I like to see if I can read the serial on the garbabe bin

http://www.steves-digicams.com/2005_reviews/5d_samples.html

[stever]

the answer is a definite MAYBE.

assuming you've got a technically good image to start with, it will still depend on the subject and type of detail in the image, viewing distance, and your expectations

most of my wildlife shots are hand-held with a 20D and 100-400, normally at f8 and ISO 400 or 800 with fill flash where possible - when printed super B (13x19) on my 2200 up-rezzed in photoshop (bicubic smoother) at 240dpi, the prints withstand close inspection which is how i like to display them - i have not been able to crop equivalent to 16x20 and get satisfactory (to me) prints

on the other hand i believe many of these images could be printed 16x20 if displayed so not to invite close inspection.  i've not played with other programs to up-rez or varying the dpi - there are vastly different opinions about up-rezzing, but i don't think there are big gains to be had one way or the other for wildlife images, but then this is a borderline situation and the sum of a bunch of small details can add up

although i think the 100-400 and 20D are well matched, better lenses will result in sharper images - my 300f4 with a 1.4x is just noticeably sharper and a 300f2.8 with 1.4x is quite noticeably sharper (subjectively i think this will make a bigger difference than the nth up-rez fiddle)

as Tim says, you'll have to try it and judge the results -- i have a touristic castle image printed super B from a Canon s400 which i would not have anticipated to be possible (and don't anticipate trying to repeat)

[Henrik Paul]

Dear Henrik,

I looked at the thread you mentioned but it seems to contain a lot of nonsensical (i.e., unrelated) mathematics and chatter. and the question asked in the inverse of my question. where I'm asking if 8 megapixels is enough to produce a sharp clear 24x18" image on my printer, the author of that thread is asking if the printer is good enough to show everything that their camera can provide.

but thank you for the reference anyway. I appreciate you're help.
[a href=\"index.php?act=findpost&pid=94127\"][{POST_SNAPBACK}][/a]

As I fail to see a meaningful difference (perhaps I just didn't understand your question right), my answer remains the same. There's pixels and then there's pixels. And then there's the endless PPI quibble. And viewing distance. And subject matter.

It depends on what your criteria for "good enough" and field of application are.