Ad
Ad
Ad
Poll
Question: megapixel
5 megapixel - 14 (3.9%)
10 megapixel - 67 (18.8%)
15 megapixel - 91 (25.6%)
20 megapixel - 60 (16.9%)
25 megapixel - 47 (13.2%)
30 megapixel - 18 (5.1%)
35 megapixel - 59 (16.6%)
Total Voters: 29

Pages: [1] 2 3 ... 5 »   Bottom of Page
Print
Author Topic: megapixel  (Read 25740 times)
BJL
Sr. Member
****
Offline Offline

Posts: 5121


« on: January 03, 2005, 05:40:15 PM »
ReplyReply

Quote
I think someday, a radical change will come about the way we capture images, tiffs will be a thing of the past.... Just for fun I'll give it a name "Fractal Unified Juxtapose Image" or FUJI for short.
No need to fantasize; replace fractals by the equally trendy mathematical concept of wavelets and you could be talking about JPEG2000. It might be five years past the advertised arrival date, but it does seem to offer a good step forward from JPEG and TIFF.

Nothing to do with pixel counts though. My poll vote was rejected becasue I read the results before voting, so I will say it here:

Somewhere between 12MP and 20MP, since 4000x5000 will certainly cover my eye's resolution needs under the closest scrutiny I have ever given to even huge prints from large format, and 3000x4000 is probably enough for that.
Logged
collum
Full Member
***
Offline Offline

Posts: 189


WWW
« Reply #1 on: January 03, 2005, 09:39:38 PM »
ReplyReply

ok :^)

48Mp of Foveon type pixels will do as well (8000x6000), on a 4x3" sensor with big pixels. i can get 4x5 quality in a 30x40 with a betterlight, so i guess i'd settle for that as well

         jim
Logged
Bobtrips
Sr. Member
****
Offline Offline

Posts: 679


« Reply #2 on: January 13, 2005, 10:57:45 AM »
ReplyReply

Quote
If the chip had some means of detecting a "filled" sensor photosite, dumping the charge, and incrementing a counter associated with that pixel, dynamic range could be increased almost arbitrarily. Each bit of the counter would add 1 stop of usable dynamic range. In addition, when the most significant bit of any counter flipped from 0 to 1, a signal could be generated to terminate the exposure to brevent clipping.
Exactly.

And - it seems to me that 'reading' the charge of an individual photosite would require using some of that charge.  Unless one had some sort of a flux meter.  (Do I have the right device?)

If you're doing a 'real time' reading then why not just send that charge on to an accumulator?  That would drain power from the p-site.  The p-site would never fill, never 'blow out'.
Logged
Andrew Teakle
Jr. Member
**
Offline Offline

Posts: 90


WWW
« Reply #3 on: January 18, 2005, 01:42:29 AM »
ReplyReply

I'd love a 36MP sensor - a 6000x6000 square sensor at 36x36mm. The projected image onto the sensor could not contain too much light falloff in the corners, but if possible... The format could be chosen from 1:1, 4:5, 3:4, 2:3, 1:2, 1:3 and 16:9. Horizontal or vertical at the push of a button. Prints could be made to 20" in long dimensions without interpolation, huge with interpolation.

While I'm dreaming, the anaolg gain (or ISO equivalent) of each pixel could be independantly adjusted to keep the image within the contrast range of the sensor i.e. while setting the exposure to keep detail in the highlights, the camera increases the ISO of the pixels recording the shadows to keep detail in the entire image.

Naturally it has an anti-static coating with ultrasonic dust removal a la Olympus, and anti-shake courtesy of Konica-Minolta. Inbuilt DxO Optic software automatically corrects for distortion, vignetting, chromatic aberration, blur, etc.

Keep dreaming!  :p
Logged

Lin Evans
Sr. Member
****
Offline Offline

Posts: 269


WWW
« Reply #4 on: February 11, 2005, 12:25:15 PM »
ReplyReply

I already have all the "pixels" I need to replace 35mm color film. What I would like is more precise autofocus, larger buffers, better color accuracy, zero metamerism, one button push manual white balance, zero moire, and no chromatic aberrations. In general, most of these wishes are achievable and will probably happen sooner if we get off the "megapixel madness" train and look for improvements in other areas. As long as there is market pressure for more pixels, we will see fewer improvements elsewhere. Actually, in the consumer venue we are beginning to see progress in this direction with the recent changes from eight to seven megapixel sensors, hybrid autofocus, lower shutter lag, etc.

In large format venues there is still no economically viable way to achieve parity in resolution with a single frame sensor capture, so until chip manufacturing processes are advanced enough to make it feasible I'll be happy with my old Kodak MF digital back.

In retrospect, some of the very best image quality I've produced have been with my Sigma SD10 (Foveon) which doesn't have a huge file size but produces exceedingly sharp and pleasing images.

Lin
Logged

Lin
Concorde-SST
Full Member
***
Offline Offline

Posts: 102


« Reply #5 on: March 07, 2005, 01:37:05 PM »
ReplyReply

hello,

well - with my 16 mp camera (1ds MkII) I have more
than enough - I dont print bigger than A3+ - and for
cropping 16 MP are perfect. Just to say it like Robert
Capa - you cant get close enough to the subject (or so)!

But - when Im allowed to dream - Id like to have
a Phase One with a full frame square sensor for
1:1 photography with a Hasselblad!

And - a full frame for the Linhof Technorama 617 !! :-)

best,

Andreas Suchert,
Concorde-SST
www.suchert.com
Logged
didger
Sr. Member
****
Offline Offline

Posts: 2030



« Reply #6 on: May 06, 2005, 11:35:12 AM »
ReplyReply

Well, the fact that this special camera shoots images that are in fact 3x the size of a normal 35mm frame goes a long way toward explaining your observations.  No one would likely claim that any 6 or 8 Mpixel digital camera is 3x as good as 35mm film.  In fact, the general claim is that 6 MP is about the same effective resolution as 35mm film but considerably "cleaner" (no grain).  8 MP is pretty much agreed to be all around superior to 35mm film.  Something like a 1ds is effectively about 2x the resolution of 35mm film, but much cleaner.  A D2X or 1dsMKII is substantially better than 1ds.  A 35mm image that's 3 frames wide is a very special case.  With my D2X I can take panorama stitching images that can go up to around 100 MPixels or more when they're all assembled together, but it's not fair to compare D2X with large format on account of this.

In any case, inspired shooting and optimal technique are more critical than the utmost in good specs for equipment.  Galen Rowell took terrific photographs with 35mm film.  However, David Muench and Michael Fatali (and many others) produce more impressively detailed large prints from large format film.  You just can't fake resolution.  You can scan for a huge amount of detail with film, but after a while you're just showing grain in very great detail.  You can uprezz and sharpen digital images, but after a while you just get a clean picture that's obviously lacking enough detail.  I doubt very very much that there's any 1ds, 1ds2, or D2X shooters that are very nostalgic about 35mm film.

Anyway, enjoy your phototgraphy, whatever side of the fence you're most comfortable on.  The digital P&S will give you a good idea whether or not you'll want to go full on digital some time.  6 MP ought to give you pretty decent 16x20.  A very successful pro commercial photographer friend of mine gets good 16x20 (depending on subject matter) from a Canon 3.4 MP P&S.  Of course, his clients only see his serious Nikon cameras and his MF and 4x5 stuff!!
Logged
williamrohr
Full Member
***
Offline Offline

Posts: 100


WWW
« Reply #7 on: June 12, 2005, 03:43:43 AM »
ReplyReply

I hope I'm not being too obtuse... but the other day I was reviewing some old B & W 16 X 20 photographs that were printed from 8 X 10 negatives  ... they were absolutely stunning.  I love my digital but even the best of the lot is not there yet.  Certainly the ease of use, etc. of todays digital is a huge plus but the overall quality is not up to the best of that "old" technology.  If you have not yet seen any of those old pictures ... check out Clyde Butcher's work as he is one of the few still practicing the "old art".  Maybe we are there (or better) if you consider Stephen Johnson's work with scanning backs but what if we could do that with a one shot back (ouch, my wallet is burning a hole in my pocket).  
Logged
Anon E. Mouse
Full Member
***
Offline Offline

Posts: 197


WWW
« Reply #8 on: August 28, 2005, 08:17:30 PM »
ReplyReply

I think there are two myths that seem prevalent in imaging. Pixel resolution (or film format) limits maximum print size. And pixel resolution equils resolving power (either with a digital camera or scanner). Both are false.

There is no limit to the size of print that can be made regardless of the number of pixels (digital) or format (film). As long as you are viewing from the correct viewing distance (the diagonal of the print), the print will appear the same no matter what size it is. The issue is the angular resolution of the eye. If you increase the viewing distance, resolution (and depth of field) appears to increase. The opposite is true if you decrease the viewing distance. So go ahead, print those images as large or as small as you like.

While there is no method of determining the resolving power of a CCD, it is not equil to the pixel resolution. It will always be something less. The resolving power is further lowered by the optics and the aperture. While the pixel resolution is fixed by the data, the ability to resolve detail is not. The joke is all all photographer whether then use a digital camera or scan their film. You are not resolving the detail that the pixel resolution suggests. Camera and scanner manufactures are quite happy not to point this out because it would cause a huge headache if consumers wanted to know the actual resolving power of their equipment.
Logged
Anon E. Mouse
Full Member
***
Offline Offline

Posts: 197


WWW
« Reply #9 on: September 01, 2005, 12:40:21 AM »
ReplyReply

Quote
Quote
Well, I would say you are simply repeating the myth. A printed image will appear the same regarless of size if the ratio of the print size and viewing distance is kept the same. It is a simple geometric problem based on the angular resolution of the eye - which is how circles of confusion and depth of field is calculated as well. That does not mean a low resolution image will not look rough, but its appearance is the same when the ratio of size to distance remains the same.
So what?

It's not a myth that if you take a 3000x2000 pixel image and a 600x400 pixel image and print them both at 16x24 and hang them on a wall side by side, the quality difference will be extremely obvious. And it's not a myth that given the fairly standard hanging-on-the-wall-at-arm's-length home/gallery viewing condition most under which most framed prints are exhibited, prints from a 1Ds or MF back can be printed larger than those from a 6MP camera and still maintain an acceptable level of image quality per unit of print area. Yes, you can get away with a lower print PPI standard as viewing distance increases, but at any given viewing distance, a file with more pixels (all else being equal) is going to make a better-looking print than one with fewer.

I am not talking about image quality vs. pixel resolution (of course more pixels make finer images). I am saying there is no limit to the size of print that can be made from a file. A bad print at 8x10 will look exactly as bad at 30x40 given the same ratio of print size to viewing distance. (Of course, if you stick your nose up to a good print, it will look bad.) This works for film; digital follows the same principles. Sharpness is based on relative size of circles of confusion. not absolute size and is defined in relation to the angular resolution of the eye.
Quote
The tests you are refering to do not specify the resolving power of the CCD (or film). Those test show the resolving power of the system - CCD (film), lens, aperture. There are tests to determine the resolving power of film, but no standard that I am aware of has been developed for CCDs or CMOS (although the manufactures must be making some assumptions to design their lenses).

Quote
While this is technically true, in practice one can measure the resolution of the sensor if one uses a lens that has a decent MTF at spatial line frequencies higher than the pixel pitch of the sensor. In that case, the sensor (or more precisely the anti-aliasing filter over the sensor, if applicable) will be the primary bottleneck limiting the resolution of the image. As noted previously, the pixel pitch of the sensor imposes a hard limit to the amount of fine detail that can be captured in an image, and this limit can be approached, but not quite reached, due to lens imperfections and the AA filter. But if the lens is good enough, the error it introduces into the measurement process is small enough that it does not significantly compromise the usefulness of the results. How do you think film manufacturers determine the MTF curve for film?
As you pointed out, IF the lens is good enough. The tests I see on the web don't seem to worry to much about the optics nor the aperture setting - seems pretty random to me. (BTW, film manufactures don't just stick any old lens on a camera and take a snap.)
Logged
Joja
Newbie
*
Offline Offline

Posts: 31


WWW
« Reply #10 on: January 02, 2005, 02:56:39 PM »
ReplyReply

With how may megapixels would you be satisfied ?
Please note, this poll only turns about the number, and not the costs !

Also think about memorycards, the storage needed in the field, resolution (both lenses and sensors), postprocessing, backing-up, hard drives, print size,

Dont think only about the number, more megapixels mean bigger files, more storage space in the field (memorycards and psds), higher quality, bigger print size without resizing, need for bigger hard drives, slower postprocessing, and the purpose (4/5, A1, prints)

Forget everething about money, the cost of the sensor, memory, hard drives, ram, software, .. Look only to the number and please, take your time to motivate your choice !!
Thank you !
Logged

------------------------------------
http://www.johanjanssens.com/
------------------------------------
Stef_T
Sr. Member
****
Offline Offline

Posts: 266


« Reply #11 on: January 03, 2005, 06:43:50 PM »
ReplyReply

I set mine down at 20MP.

The question after a long time becomes print size as well. Do you really plan on making a billboard print? Because if you're not, it's pointless to have 1000ppi on a 19*13 print. For me 15MP would be ideal, to have 19*13 prints at about 300ppi, true this is overkill, but it would make a great image, that would be better then 200ppi under extreme scrutiny. It would also allow for a larger 24*19 print, if I so wished. 20Mp was chose, to include a crop.

Obviously if the cost was a non-issue, I'd most likely get higher, but if the cost was another 2,000$ for 25Mp, then no chance of me getting it.

Stefan
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8854


« Reply #12 on: January 03, 2005, 09:31:21 PM »
ReplyReply

Cor blimey! It's not just the number that counts but the quality. Always remember, it's system[/i] resolution that's important; the combination between lens and sensor. We're after a 'system' MTF response. The higher, the better at any given resolution.
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5121


« Reply #13 on: January 13, 2005, 11:02:26 AM »
ReplyReply

The photosites do have a way of detecting nearly full pixels mid-way through the exposure, and A/D conversion is done right at each photosite, according to comments at the Pixim site. Here is my speculation about how that technology might work, based or reading the Pixim web page and some earlier comments on the research at Stanford U.

For an exposure of say 1/125, each photosite checks the electron count in each of its well after say 1/8000s. Wells that are close to full are then read, converted to a digital value right there at the photosite, and multiplied by 64 since the exposure was cut of after only 1/64th of the full time. Multiplying by 64 (or any power of 2) is easy: shift the bits six places to the left.

The same is done again after 1/4000s, with photosites read then having the value multiplied by 32 (shift five bits left).

Repeat at 1/2000, 1/1000, 1/500, 1/250, 1/125. Digitised values are then read off the sensor all together after the exposure is finished.

Effectively, the sensor reads one stop of the luminance range at a time, from the top down. This would have another bonus: even with a fairly long total exposure time, the highlights to mid-tones are read over a shorter time, so motion blur mostly comes up down in the shadows; twice as bad for every stop darker.

If you want, you could examine an image and set the black point at a level that cuts off any excessive shadow blur. But that might rarely be needed, since the blur could be so dim as to cause few problems.
Logged
61Dynamic
Sr. Member
****
Offline Offline

Posts: 1442


WWW
« Reply #14 on: January 17, 2005, 11:16:27 PM »
ReplyReply

I'm more interested in optics than megapixels, however megapixels do come in usefull in post and for resolving detail.

Assuming optics are not the problem they are today...
-Landscape/portrait/still images: anything around 25-30MP in a full-frame medium-format sensor.
-General-use SLR: 10MP
- Candids/street photography: I'd be happy with a full-frame 24x36mm sensor made to be very capable in low-light photography (ISO 6400 please). What MP rating? 4MP would be just dandy. All wrapped in a RF of course.

And I'm going to continue my push for modular design so that I can interchange between color, true BW, and IR-only sensors.
Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #15 on: February 08, 2005, 12:22:00 PM »
ReplyReply

We'll probably live to see that being affordable. 10 years ago the idea of the 1Ds was pretty radical and outrageous.
Logged

Dwight Arthur
Newbie
*
Offline Offline

Posts: 13


« Reply #16 on: March 17, 2005, 11:59:08 PM »
ReplyReply

I agree with those who point out that lenses must be able to produce adequate resolution to take advantage of greater sensor resolution. The other issue is crafstmanship: you really have to do everything else right to even get the most out of a great lens. I use a 1Ds, and find that often it is my unwillingness to use mirror lockup, for example, that is the real limiting factor in image sharpness, and a 16 MP sensor would then produce no better an image anyway! I don't think many people, pros included, really do all they can to use the resolution they currently have, even as they dream about more! But for those who do, 25 megapixels would be a minimum if they don't need to crop, and love to work with big display prints, I would suppose.
Logged
DAL
Newbie
*
Offline Offline

Posts: 4


« Reply #17 on: May 05, 2005, 12:20:50 PM »
ReplyReply

Hey didger, I'm kind of new to this so take it for what it's worth  . He used a hasselblad Xpan camera (first time I've heard of it) which uses 35mm transparency film, in this case fuji velvia or maybe it was Provia, I'm going back to see it again, in which the camera exposes about 3 frames worth or whatever a 6x7 format is, so you get long narrrow transparencies. I had a chance to talk to the photographer and he said he had worked with or under Graham Nash? from Crosby Steels and Nash?(the band). He said he used the best scanner, best available paper and inks. He also said he exposed the film using the film 'reciprocity exposure' method? Anyhow color was magnificent, sharpness and detail was outstanding,and the photographer has a great eye. The showing is in Toronto Canada, if anyone with more experience than me would like to check it out. Like I said it blew me away and if someone 'who knows better' can give a better analysis it would be much appreciated because personally 'I' have never seen pictures this vibrant or sharp. I have dabbled in photography for about 15 yrs, belonged to a camera club and won some 'minor' awards. I use a T90, OM4T and various quality lenses. When I shoot 'for real' I always use transparency films. I would love to sell all my gear and go digital, and almost did, but after veiwing these photo's I've jumped back to the other side of the fence for now . I have just bought a point & shoot, 6mp, a few days ago and have been amazed at the 'convenience' it provides. My first 11x14 or 16x20 may push me back over the fence.
Don
Logged
Bobtrips
Sr. Member
****
Offline Offline

Posts: 679


« Reply #18 on: June 12, 2005, 12:17:41 PM »
ReplyReply

Quote
I hope I'm not being too obtuse... but the other day I was reviewing some old B & W 16 X 20 photographs that were printed from 8 X 10 negatives  ... they were absolutely stunning.  I love my digital but even the best of the lot is not there yet.
Well, it's not that easy to carry an 8 x 10 into the field, set it up, and shoot it.

So how about carrying a 1Ds MkII instead, setting it up vertically and taking the scenes in two frames?

Then you get a 4992 pixel by 6000+ pixel file once stitched.  (You loose a bit of the 3328 x 2 sides when overlapping for alignment.)

That will allow you to print a 16" x 20" at 300 PPI.  And since most of us (apparently) can't resolve much above 250 PPI then you've met your LF film needs digitally.

(Or you could use a more affordable digital and shoot more frames and do a bit more stitching.  Probably a lot less work than shooting huge hunks of film.)

(Oops, saw the B&W after writing the above.  Gonna need to shoot more fames to equal fine grain B&W, but the technique still holds.  Or does that hold if one is only making contact prints?  Maybe I was right anyway....)
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8854


« Reply #19 on: August 21, 2005, 10:25:25 PM »
ReplyReply

Quote
Not silly, actually. I shoot to print, which is the most demanding application in terms of pixels. The largest print I would ever want is 20x24.
Fair enough! It's good to know there are some people who know exactly what they want and don't want  Cheesy . However, if you wanted a 20x24 print from just a part of an image, the part that was compositionally interesting, then greater than 39MP might have been useful.

The sensible answer to the question, 'how many pixels do you want?', is:-  "Sufficient to extract the full potential of any lens I might use with the camera."
Logged
Pages: [1] 2 3 ... 5 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad