Pages: 1 [2] 3   Go Down

Author Topic: Pixelsize and depth-of-field  (Read 36640 times)

Graham Mitchell

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2281
Pixelsize and depth-of-field
« Reply #20 on: May 25, 2008, 08:23:19 pm »

Here is a 100% crop from the 22MP image, where you can clearly see the areas of the image which are in and out of focus.

[attachment=6772:attachment]
Logged

Brent Daniels

  • Full Member
  • ***
  • Offline Offline
  • Posts: 129
Pixelsize and depth-of-field
« Reply #21 on: May 25, 2008, 09:34:25 pm »

I think this discussion is getting a bit confused. What I think Henrik is referring to is image sharpness lost due to small aperture diffraction not depth of field, which are two different things.

Does pixel size, density, and design have an effect on the diffraction effects (softening) of small lens apertures required for maximum DOF ?
 
- Why with digital it seems the diffraction effects that shows when stopped down to f-22 to obtain the required DOF the critical focused point of an image does not look as sharp as the would be at f8 compared with past film experiences?
- Why with digital when one applies camera movements you seem to lose more sharpness at the critical focused point compared with no movements, as compared to past film experiences?

One non pixel related answer is that even with film there was such effects. We only inspected our transparencies with 10x magnification, and not at the type of magnification digital allows.

At present there would seem to be a direct relationship between pixel size, density in the chip, and pixel design in images degrading in sharpness due to diffraction with smaller apertures and or with technical camera movements. If the pixels properties were not an issue at all the P31+ and Blads 31mpx chips would work great with wide angle lens and technical camera movements.

I know I had less effects from camera movements with my older Sinar H23 back as compared to the Blad 528C back.

I have however seen examples, and tested it to be correct, that if you critically focus on one point wide open and then stop down and shoot, this point will not be as sharp as if you refocus on the exact same point at f22. Is this an effect of digital pixels , or a lens effect? I never tried it with film so I have no idea.

The big question, and Henrik's question is if real life research / testing has been done and if there are any results available to the public?

Brent Daniels
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Pixelsize and depth-of-field
« Reply #22 on: May 25, 2008, 09:58:53 pm »

I would venture to suggest, at the risk of getting vilified, that having fewer but larger pixels on the same size sensor can contribute to an appearance of greater DoF, depending on print size and viewing distance.

As Graham has suggested, downsizing the image which has more pixels to the same size as the image with fewer pixels, sort of equalizes DoF in both images.

But what happens if you upsize the image with fewer pixels, through interpolation, to the same size as the other image? In those circumstances, the image that started off with fewer pixels might appear to have greater DoF because the sharpness at the plane of focus will be less (than that in the image with initially more pixels) and will be closer to the sharpness of the rest of the image (in both images) which is not in the plane of focus.

Downsizing the image which has the greater number of pixels results in a discarding of resolution at the plane of focus, but no discarding, or less discarding of resolution away from the plane of focus.

Upsizing the image with fewer pixels is probably a fairer comparison because no information is thrown away in either image.

However, there might be a problem with the definition of DoF if it includes phrases such as 'acceptable sharpness'. To take an extreme example, a large print from a pin-hole camera might have amazing DoF, everything being equally sharp (or fuzzy) from one's feet to infinity. Pin-hole cameras are noted for their great DoF.

If nothing in the print is acceptably sharp, does that have a bearing on DoF? If you were to take a slightly fuzzy, image from a pinhole camera, digitise it and then through photoshop manipulation, seamlessly plant a sharp object ( figure, face, tree, whatever) in the middle of the image (or replace a relatively small object already in the image with a sharper version), would we have converted an image that we previously considered to have great (extensive) DoF into one which now has shallow DoF?
« Last Edit: May 25, 2008, 10:04:56 pm by Ray »
Logged

newrooky

  • Newbie
  • *
  • Offline Offline
  • Posts: 21
    • http://www.brandonwaterman.com
Pixelsize and depth-of-field
« Reply #23 on: May 25, 2008, 10:29:34 pm »

Quote
It is unhealty for a discussion to mix up different aspects. Stick to narrow subjects, abstracted from others; otherwise we end up with arguments like my eyes are not as good as yours, so the DoF for me is much larger than for you.
Well, then back to the basics.
...
Thie means with sensors, that as long as the light rays arrive within the "catchment area of the pixel site" (i.e. over the microlens of a single pixel), then one can not differentiate between in-focus and out-of-focus. However, if some of the light rays arrive at other pixels, then that point of light appears to be at several pixels at the same time.

So, the larger the pixel site, the larger difference is tolerable.
[a href=\"index.php?act=findpost&pid=197968\"][{POST_SNAPBACK}][/a]

As you mention, this topic has, like almost all, become (in great part do to myself, sorry) a multi-faceted conversation.  So yeah, this will be my last response and if anyone does care to discuss the effects of pixel density on DOF further, please message me.  Whether i be right or wrong I am always up for learning and listening, especially when i am wrong.

In regards to larger pixels taking multiple rays of light and therefore decreasing dof:  While the gradient between in focus and out of focus will not be as smooth on the higher mpx chip, the amount between in focus and out of focus will remain the same.  The DOF is inversely proportional to sensor size and, physics wise, doesn't have anything to do with pixel density.  The only difference is in smoothness of in/out of focus gradation.

thanks for your replies and such,

-b-
Logged
[span style='font-family:Geneva'][span style='color:gray']Brandon Waterman   -  Photo+[/span][/span]

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Pixelsize and depth-of-field
« Reply #24 on: May 25, 2008, 11:34:20 pm »

Quote
In regards to larger pixels taking multiple rays of light and therefore decreasing dof:
The larger pixel size increases the DoF.

Quote
The DOF is inversely proportional to sensor size
No relationship whatsoever.
Logged
Gabor

newrooky

  • Newbie
  • *
  • Offline Offline
  • Posts: 21
    • http://www.brandonwaterman.com
Pixelsize and depth-of-field
« Reply #25 on: May 26, 2008, 12:04:45 am »

Quote
The larger pixel size increases the DoF.
No relationship whatsoever.
[a href=\"index.php?act=findpost&pid=197986\"][{POST_SNAPBACK}][/a]


ok, within the limitation to them having the same fov this is true.

"More precisely, if photographs with the same final-image size are taken in two different camera formats at the same subject distance with the same field of view and f-number, the DOF is, to a first approximation, inversely proportional to the format size. "

makes sense to me, no?
Logged
[span style='font-family:Geneva'][span style='color:gray']Brandon Waterman   -  Photo+[/span][/span]

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Pixelsize and depth-of-field
« Reply #26 on: May 26, 2008, 12:12:31 am »

Quote
ok, within the limitation to them having the same fov this is true.

"More precisely, if photographs with the same final-image size are taken in two different camera formats at the same subject distance with the same field of view and f-number, the DOF is, to a first approximation, inversely proportional to the format size. "

makes sense to me, no?
[a href=\"index.php?act=findpost&pid=197990\"][{POST_SNAPBACK}][/a]

Makes sense to me also, but in digital imaging the size of the unaltered image depends on the number of pixels not the size of the pixels nor the size of the sensor. So in my view, the different number of pixels on two sensors of the same size can affect the appearance of DoF, depending on how those two images are processed and what size prints are produced.

Lens sharpness will also have a significant role to play. If the lens used with both cameras of equal sensor size but unequal pixel count are simply not good enough to deliver that additional resolution that the sensor of higher pixel count is potentially capable of, then DoF differences would probably be negligible.

In any case, such differences due to pixel count would be a secondary concern. However, it would be an interesting experiment to make if one had nothing better to do. I would think that comparing differences of less than double the pixel count might be a waste of time. Comparing a D30 with a 40D, or a D60 with a 450D, or a 1Ds with 1Ds3 might produce some worthwhile differences in DoF, using both high quality primes and low quality zooms.
« Last Edit: May 26, 2008, 01:16:31 am by Ray »
Logged

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Pixelsize and depth-of-field
« Reply #27 on: May 26, 2008, 01:00:47 am »

We are talking digital imaging.

A digital image is a digital image is a digital image... not a piece of paper, not a monotor, nothing upresed or downresed. A digital image consists of pixels, and that's it.

A point of light is out of focus if its image appears in several pixels. This has nothing to do with feld of view, sensor size, print size, etc. Otherwise the entire discussion is total nonsense. You print with 360ppi, I print with 300ppi. You look at it from far, I look at it from close. Your eyes are good, mine are bad. You denoise it, I don't. Your monitor has a 72dpi, my one has 96dpi. And so on.
Logged
Gabor

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Pixelsize and depth-of-field
« Reply #28 on: May 26, 2008, 01:40:42 am »

Quote
We are talking digital imaging.

A digital image is a digital image is a digital image... not a piece of paper, not a monotor, nothing upresed or downresed. A digital image consists of pixels, and that's it.

[a href=\"index.php?act=findpost&pid=197994\"][{POST_SNAPBACK}][/a]

Gabor,
Whilst all my digital images exist in some file format of one sort or another, DoF is only apparent on a monitor or on a print after the file has undergone some degree of processing.

Having more pixels without changing the sensor size can never reduce resolution in any part of the image, but it can increase resolution in the plane of focus.

I take two shots of a face with the same high quality lens, one with a 3mp D30 and the other with the 10mp 40D. I use an aperture such that the nose is perceptibly OoF in the 40D shot but the eyes are exactly in focus.

I take the same shot with the D30, focussing on the eyes so that the nose is perceptibly OoF, just as in the 40D shot.

I uprezz the D30 image to the same size as the 40D image and make equal size prints. I view both prints from as close as I like. What do I see?

I see the nose in both shots as being equally sharp, but the eys in the 40D shot will be sharper (10mp at the plane of focus has to be sharper than 3mp, with a good lens, or what's the point?).

In the 10mp 40D shot, I get a sense of a slightly shallow DoF (because the nose is less sharp than the eys). In the 3mp D30 shot there is no sense of a shallow DoF. The eyes and nose are both equally sharp.
Logged

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA
Pixelsize and depth-of-field
« Reply #29 on: May 26, 2008, 02:05:17 am »

The sensor well size does affect DOF in that the use of small apertures is limited by smaller sensor wells.  In other words the diffraction is noticeable sooner the smaller the well.  The new canon 1Ds3 for example will probably not be as good for macro work as the original 1D since you can stop down to f/11 or f/16 before really noticing the diffraction while you see the diffraction with the 1Ds3 at f/6.8 - so because you can stop down further you can get more apparent DOF.
Logged
Rolleiflex USA

Sheldon N

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 828
Pixelsize and depth-of-field
« Reply #30 on: May 26, 2008, 02:09:08 am »

Increased pixel size means lower resolution for any given format/sensor size. Lower resolution, when viewed at 100% magnification, is the equvilalent of a smaller print size in comparison to a higher resolution imaging device.

Since larger pixel size means lower resolution (for a given format size), YES it will be less susceptible to diffraction and will allow for a smaller f stop before diffraction negatively affects the resolution of the imaging system. HOWEVER, a smaller pixel size can produce the same resolution at the same f stop, but can also produce higher resolution when shot at more wide open f stops. Diffraction limited is diffraction limited, regardless of how big your pixel size is.

The applicable COC is determined by enlargement factor. Enlargement factor is determined by how big your original format size is relative to your final print size. So, the things that really matter for DOF are 1) Format/Sensor size.   2) Final print size.  3) Focal length.  4) Aperture.  5) Focus Distance.  

If your DOF standard is how the image looks at 100%, then pixel size matters - primarily because larger pixel size means less resolution, and less resolution means a smaller effective print size (if your print size is 100% view). That smaller print size means more DOF.

If you equalize the resolution (same megapixels) but compare larger vs. smaller pixel sizes, you mess with all the variables, because now you have different format sizes which require different focal lengths. The longer focal length required by the larger format will lead to a smaller DOF for the larger format, hence the "rule" that larger formats have less DOF.

If you use the same format size and compare large pixels vs. small pixels (low res vs. higher res), then you really need to compare similar print sizes. This means not printing the higher res image at its maximum print size but limiting it to the resolution of the low res/large pixel system. Given equal print sizes, the DOF will be the same. If you increase print size, the DOF of the high resolution/small pixel system will be less. However, this is because of the larger print size, not because of the smaller pixels.

Anyhow, all this said, larger pixel size does not gain you anything other than a resolution limitation, or forcing you into a larger format size.
Logged
Sheldon Nalos
[url=http://www.flickr.com

thsinar

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2066
    • http://www.sinarcameras.com
Pixelsize and depth-of-field
« Reply #31 on: May 26, 2008, 02:33:34 am »

... and I would like to add, on the "observing/viewing distance" of the final output.

Thierry

Quote
The DoF depends on the focal length, focusing distance, aperture diameter and circle of confusion. The CoC depends on the pixel size.
[a href=\"index.php?act=findpost&pid=197946\"][{POST_SNAPBACK}][/a]
Logged
Thierry Hagenauer
thasia_cn@yahoo.com

Frank Doorhof

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1522
    • http://
Pixelsize and depth-of-field
« Reply #32 on: May 26, 2008, 05:09:58 am »

As Thierry mentioned it's all in the viewing distances.

First off all DOF is determined by the lens you are using.
That's why you will have much less DOF on a MF system than on a DSLR and even more on a P&S.

Although you will have a similair scene (although they will differ in FOV) the MF will have less DOF than the FF DSLR which will have less than the Crop DSLR etc.
This is mainly due to the difference in lens.
On the MF you will use a 80mm on the FF DSLR a 50mm on the crop a 32mm.

Now take into consideration the Circle of Confusion.
This is connected to the printsize and viewing distance.

The crop that is shown does show less DOF than the full picture, but that's logical because the crop is the same size as the whole downsized picture, meaning the printsize (the pixels) has gone up but the viewing distance is the same (your eyes to the monitor).

For me this was the reason to switch to MF, I love to work with shallow DOF on some occasions and with the DSLR system I had to switch to long lenses or ridicilous large apertures.
With the MF system I can just drop the ISO to 25 and use smaller apertures and still get the same results.
In that case it's all in the longer lenses you can use without changing the distance to the model.
Logged

henrikfoto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 899
Pixelsize and depth-of-field
« Reply #33 on: May 26, 2008, 05:14:47 pm »

Quote
The sensor well size does affect DOF in that the use of small apertures is limited by smaller sensor wells.  In other words the diffraction is noticeable sooner the smaller the well.  The new canon 1Ds3 for example will probably not be as good for macro work as the original 1D since you can stop down to f/11 or f/16 before really noticing the diffraction while you see the diffraction with the 1Ds3 at f/6.8 - so because you can stop down further you can get more apparent DOF.
[a href=\"index.php?act=findpost&pid=198002\"][{POST_SNAPBACK}][/a]


Thank you, Eric! This was really what my question was about. All the newer digital lenses (like the Apo Sironar Digital HR lenses) are best at f8 or close to that. When you go to smaller appertures the lenses are not as good. As I undersand these lenses are made like that because the smaller pixels of the modern sensors don´t work that well with these small appertures. As a result of this we have less depth-of-field because the use of f-stops above f16 (or even lower) are limited.

Anybody agree on this?
Logged

Nick Rains

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 705
    • http://www.nickrains.com
Pixelsize and depth-of-field
« Reply #34 on: May 26, 2008, 06:31:26 pm »

Quote
Example. At full size, the DOF is only half a metre. At this size, the whole image looks in focus.


[a href=\"index.php?act=findpost&pid=197930\"][{POST_SNAPBACK}][/a]

The smaller the image is viewed, or the further away, the more apparent depth of field. It's all to do with COCs, not sensor or pixel size in this case.
Logged
Nick Rains
Australian Photographer Leica

Graham Mitchell

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2281
Pixelsize and depth-of-field
« Reply #35 on: May 26, 2008, 06:38:19 pm »

Quote
The smaller the image is viewed, or the further away, the more apparent depth of field. It's all to do with COCs, not sensor or pixel size in this case.
[a href=\"index.php?act=findpost&pid=198162\"][{POST_SNAPBACK}][/a]

No matter how much large you print that image or how closely you view it, the even focus will still be there.

It is complicated by the fact that there are several limiting factors. If the resolution is high enough, then viewing distance and print size comes into play.

Perhaps it would  be more accurate to say that for a given print size and viewing distance, there will be a pixel size limit above which the DOF will visibly widen.
Logged

jjj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4728
    • http://www.futtfuttfuttphotography.com
Pixelsize and depth-of-field
« Reply #36 on: May 26, 2008, 07:04:48 pm »

I'm with Nick, your reducing the size of image to a lower MP size is a red herring in the context of OP's question. Effectively you've moved yourself further away from the image as it's now smaller and that is why you now cannot percieve the OoF aspects of image. DoF is determined not only when you take photo but how you view it. Acceptable DoF depends on how big/small you print/display image and how far away you view it.
Plus with downsizing, the sharp parts of image are degraded so there is less contrast between them and the originally much less sharper parts of image.
Logged
Tradition is the Backbone of the Spinele

henrikfoto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 899
Pixelsize and depth-of-field
« Reply #37 on: May 26, 2008, 07:10:33 pm »

But this is all aoutside the topic. The question was the relationship between the sensors pixelsize and possibility to use smaller apperture, and thereby increased DOF.
Logged

Gordon Buck

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 458
    • LightDescription
Pixelsize and depth-of-field
« Reply #38 on: May 26, 2008, 07:33:55 pm »

Quote
Is it right to assume that the optimum setup to increase depth-of-field would be a sensor with big pixels?
[a href=\"index.php?act=findpost&pid=197924\"][{POST_SNAPBACK}][/a]

At dofmaster.com, the online calculator produces the same results for hyperfocal distance and depth of field for the Canon 20D, 30D, 40D which have the same size sensor but increasing number of pixels.  Likewise, calculations for the Canon G series digicam from G1 to G9 produce the same results even though the number of pixels in the sensor has increased over the years.  In these calculations, the Circle of Confusion is assigned according to some logic or other.

Now, I don't know enough about the details to know if the calculations are correct but I do know enough about DOF and Circle of Confusion to realize that this is a highly discussed and debated topic.  It seems to me that the classic calculation is based on enlargement and not number of pixels.  In the good old days, the depth of field scale on a lens did not vary with the film speed or resolution.

So my bet is that, to the first approximation, the amount of enlargement -- as related to sensor size -- is the most significant parameter.
Logged
Gordon
 [url=http://lightdescription.blog

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Pixelsize and depth-of-field
« Reply #39 on: May 26, 2008, 09:12:58 pm »

Quote
At dofmaster.com, the online calculator produces the same results for hyperfocal distance and depth of field for the Canon 20D, 30D, 40D which have the same size sensor but increasing number of pixels.  Likewise, calculations for the Canon G series digicam from G1 to G9 produce the same results even though the number of pixels in the sensor has increased over the years.  In these calculations, the Circle of Confusion is assigned according to some logic or other.

Now, I don't know enough about the details to know if the calculations are correct but I do know enough about DOF and Circle of Confusion to realize that this is a highly discussed and debated topic.  It seems to me that the classic calculation is based on enlargement and not number of pixels.  In the good old days, the depth of field scale on a lens did not vary with the film speed or resolution.

So my bet is that, to the first approximation, the amount of enlargement -- as related to sensor size -- is the most significant parameter.
[a href=\"index.php?act=findpost&pid=198184\"][{POST_SNAPBACK}][/a]

I can't see what the confusion is here. People buy cameras with increasingly greater pixel count because they hope to get higher resolution. You can't get higher resolution at small apertures where the lens is limited by diffraction. The higher the pixel count (on the same size sensor), the better the lens needs to be in order to achieve that higher resolution.

Having achieved your higher resolution, by either stopping up to a bigger aperture which is less affected by diffraction, or by using a better lens, a print from the image with the higher resolution must have a shallower DoF than the same size print from the lower resolution image, assuming the same lens was used at the same aperture and same distance to subject.
Logged
Pages: 1 [2] 3   Go Up