Luminous Landscape Forum
Raw & Post Processing, Printing => Printing: Printers, Papers and Inks => Topic started by: torger on January 23, 2015, 06:38:39 am
-
So there's capture sharpening and there's output sharpening. Let's leave creative sharpening for a while.
Capture sharpening is to sharpen up the capture image, ie pixel peep sharpening in the raw converter. Then the image is scaled, usually up, to match printer resolution, and then "output sharpening" is applied to compensate for smearing of the print/paper.
To me it seems something is broken with this, or I haven't understood how it works. The way I think it should work is that no sharpening is applied until you've scaled up the image.
If you like me have a very sharp camera without AA filter you get jaggies, and if I do capture sharpening first I enhance those jaggies. I really dislike visible jaggies in prints. I'm thinking that if instead USM sharpening is applied after bicubic scaling up (USM parameters scaled to match of course!), you could get the same amount of sharpening with less risk of staircasing. On top of that you would add any output sharpening.
But maybe this is already how say Lightroom works when you generate a print, ie all sharpening job is made after bicubic scaling, or is it a two step phase?
I just got my own printer and is figuring out workflows, so I'd like to know how this works (or should work).
-
As I understand it , the capture sharpening is to get rid of the effects of the AA filter (which blurs thing slightly) so presumably if you don't have an AA filter, then there is no need for capture sharpening.
And output sharpening depends on the resolution used at print, and the type of paper.
keith
-
A nice tutorial if you haven't already reviewed can be found here:
http://www.cambridgeincolour.com/tutorials/image-sharpening.htm (http://www.cambridgeincolour.com/tutorials/image-sharpening.htm)
I think the goal of capture sharpening is to apply only so much sharpening as can be done with no visible detrimental effect to the image. That said, if your original image contains defects (eg. jaggies), then you're simply faced with a creative choice of whether to enhance sharpness of the non-defective areas, at expense of the defective area... or not. Basically I'd say jaggies would be equivalent to finding out after the fact, that there's an element you captured in your scene that you didn't want.
Alternatively sharpen away and then use eg. a smoothing/blur brush on the jaggies areas.
-
There can be jaggies and a need for sharpening at the same time. Consider the following 100% crop of a silly test shot I made with my camera (H4D-50 mounted on Linhof Techno with a 72mm Schneider Digitar lens, can't remember if it's an f/11 or f/16 shot, probably f/11). The bins shows some colorful moiré too, but lets leave that out for a while. Everything is a little soft from diffraction and demosaicing, so I would normally do capture sharpening on this. But if I print big the resulting staircasing in the straws can be an issue.
I haven't really had time to try yet, but as desribed in the intitial post it seems to me to be wiser to scale up for final print size first and then sharpen with a larger radius, rather than first sharpen at the original size then scale up and sharpen again for output.
Edit: some brief testing seems that it might be an advantage to do it in a single pass in the end, but it's a lot about precise tuning of parameters to make it work and really make a difference. The further blurring of the upscale needs to be taken into account too. If everyone's doing it two-pass maybe I should too...
-
So there's capture sharpening and there's output sharpening. Let's leave creative sharpening for a while.
Capture sharpening is to sharpen up the capture image, ie pixel peep sharpening in the raw converter. Then the image is scaled, usually up, to match printer resolution, and then "output sharpening" is applied to compensate for smearing of the print/paper.
To me it seems something is broken with this, or I haven't understood how it works. The way I think it should work is that no sharpening is applied until you've scaled up the image.
If you like me have a very sharp camera without AA filter you get jaggies, and if I do capture sharpening first I enhance those jaggies.
Hi,
Well, with Capture sharpening you actually attempt to restore the input signal (the original scene) that the improperly low-pass filtered sensor captured, by removing optical (=residual lens aberration, diffraction) blur, thus revealing all ugliness that a non-AA filtered image stands for. You are, if done well, not enhancing jaggies that are already there, you are just removing some blur that sort of kind of hid it a bit from plain sight.
You will exaggerate the visibility of jaggies from non-OLP filtered image captures when the Capture sharpening parameters are wrong (typically too large a radius with too large an amount).
I really dislike visible jaggies in prints. I'm thinking that if instead USM sharpening is applied after bicubic scaling up (USM parameters scaled to match of course!), you could get the same amount of sharpening with less risk of staircasing. On top of that you would add any output sharpening.
It's not too far fetched, I've proposed it as an alternative workflow before, but it is technically (slightly) inferior to specifically addressing the optical blur (+demosaicing), and specifically addressing the upsampling+media/viewing distance compensation. When all source and processing blurs are lumped together in a blur-on-blur cascade, the resulting blur will be kind of Gaussian in nature. Gaussian blur is an excellent candidate for deconvolution sharpening, and e.g. FocusMagic does quite a nice job, also if only used on the final output file size.
The quality of the upsampling algorithm therefore also becomes more important. Sharpening of upsampling artifacts will also look ugly. So if one uses a workflow with postponed Capture sharpening + Output sharpening, a low artifact generating algorithm will allow to really push the final sharpening. Topaz Labs Detail (http://www.topazlabs.com/detail) is another amazing tool, also for output sharpening. It has deconvolution sharpening capabilities (to address some of the postponed Capture sharpening), and allows to (halo free) enhance (increase/reduce) the amplitudes of different detail levels. When used as an output sharpening tool, the smaller detail controls are more useful than when used as a Creative 'sharpening' tool where all detail levels can benefit from the various detail size controls.
Topaz Labs plugins can be directly accessed (https://support.topazlabs.com/hc/en-us/articles/200091347-Accessing-Topaz-Labs-in-Lightroom) from Lightroom after installing the free Fusion Express plugin. So you could export an output size ready but unsharpened file, and send that to e.g. Detail for the finishing touches without having to leave LR.
But maybe this is already how say Lightroom works when you generate a print, ie all sharpening job is made after bicubic scaling, or is it a two step phase?
It's hard to say what Lightroom exactly does under the hood, but since sharpening and noise reduction are apparently linked, I would guess that the output sharpening step is a separate/cumulative operation based on pre-sharpened data.
Cheers,
Bart
-
Thanks for the detailed feedbacks. I shall try to not overdo this too much, but optimizing these things I also think is kind of fun... I'll experiment some and see what I get.
I don't think being without AA filter on the camera is a great idea, but in the MFD world that's the only option you have so that is what I have to get to work as well as possible.
I have noticed that I have a different taste on what looks good in terms of fine details, and might therefore want an unconventional workflow. When it comes to aliasing people in general doesn't seem to care, or rather the opposite if they don't see aliasing the lens is not sharp enough or there's too much diffraction. That's why D800E was more popular than D800, despite that I'd say the D800 (with AA filter) produces better looking fine details. Likewise upscalers like perfect resize is sharp but painterly looking. I want a smooth look that is as sharp as possible up close, but not showing any sign of that it's a digital picture consisting of pixels.
-
Topaz Labs plugins can be directly accessed (https://support.topazlabs.com/hc/en-us/articles/200091347-Accessing-Topaz-Labs-in-Lightroom) from Lightroom after installing the free Fusion Express plugin. So you could export an output size ready but unsharpened file, and send that to e.g. Detail for the finishing touches without having to leave LR.
It's hard to say what Lightroom exactly does under the hood, but since sharpening and noise reduction are apparently linked, I would guess that the output sharpening step is a separate/cumulative operation based on pre-sharpened data.
Bart,
I do have Fusion Express installed as an external editor for Lightroom. Since Fusion Express requires saving a TIFF, I see no advantage in using Photoshop as the external editor and accessing the Topaz plugins from Photoshop. Fusion Express would be helpful for those who don't use Photoshop, but is there any advantage in using Fusion Express over Photoshop as an external editor. I often use Topaz Clarity and other Topaz plugins together in succession as well as other adjustments in Photoshop and find using PS as the external editor is more convenient. Am I missing something.
Thanks,
Bill
-
I think Bruce Frasier was one of the ones that started the 2 pass sharpening method.
http://www.amazon.com/World-Sharpening-Photoshop-Camera-Lightroom/dp/0321637550/ref=sr_1_1?ie=UTF8&qid=1422039802&sr=8-1&keywords=real+world+sharpening
-
Bart,
I do have Fusion Express installed as an external editor for Lightroom. Since Fusion Express requires saving a TIFF, I see no advantage in using Photoshop as the external editor and accessing the Topaz plugins from Photoshop. Fusion Express would be helpful for those who don't use Photoshop, but is there any advantage in using Fusion Express over Photoshop as an external editor.
Hi Bill,
Fusion Express allows to right click an image and use "Edit in". There is no immediate advantage over using Photoshop as the host program for the plugin that I can think of at this moment, other than not requiring Photoshop or other Photoshop plugin enabled image editor. The resulting enhanced TIFF can then be added to the LR catalog, or discarded after use. The only possible benefit could be that less memory is required while keeping Lightroom open, without needing both LR and PS open at the same time (to allow switching between them).
I often use Topaz Clarity and other Topaz plugins together in succession as well as other adjustments in Photoshop and find using PS as the external editor is more convenient. Am I missing something.
I don't think you are missing anything, in fact by using the plugins in Photoshop one adds the possibility of masked layers that can later be altered. Even more powerful is the ability to convert a duplicate layer to a Smart Object before using one or more Topaz Labs plugins on that layer, which allow to revisit all the plugin's settings at a later stage. That's particularly useful when more layers are added later in the workflow, after the Topaz plugins have been applied early on.
Cheers,
Bart
-
If you're camera is as sharp as you claim and is somewhat a recent model I'm assuming you're working with at least 16MP.
At this high resolution I'ld suggest you examine the size of the each pixelated stair step and compare it to the size of the ink droplets at the final sharpened or non-sharpened print size.
I print with a 6MP DSLR with a very thin AA filter which does produce on occasion heavy stair stepping along straight edges, but even at that resolution I can't see the stair steps due to the inkjet's stochastic fine droplet pattern. And I'm printing at 100% size (no upsampling) on a cheap Epson "All In One" on glossy paper with much larger ink droplets than a pro line model.
All you have to do is run a small print test to see if the stair step edges can really be seen viewing the print at normal distance.
-
Hi,
No, there is a lot of loss of modulation weather OLP filter is used or not. There is an MTF of the lens which should be less than 10% (or so) to avoid aliasing. A modern sensor often resolves at around around 80 lp/mm. MTF data seldom go that far, but I would guess that only the very best lenses reach say 30% at 80 lp/mm.Normal lenses may achieve far less. Image below is for Zeiss Planar 80/2.8 at f/8 near optical axis on P45+.
(http://echophoto.dnsalias.net/ekr/Articles/Aliasing2/Test3/Plots/NoSharpening.png)
Now, the sensor itself is subject to diffraction. So the sensor does also have an MTF well below 100%.
To that comes diffraction which is significant stopped below f/8 or so.
So, capture sharpening needs to compensate for all that.
Now, as Anders points out, any lens that has significant MTF at Nyquist will cause severe aliasing. Capture sharpening also enhances aliases. On the other hand, if the image is upsized the artefacts will also be enlarged before enhancement.
As a small comment, I am just in progress of making an exhibition print from an 10 MP APS-C image. I happen to have a 24 MP image of the same subject, but cloud patterns are better on the 10 MP image. In this case I made two different prints:
- An A2 size print from an unsharpened image, upsized in Photoshop to 574 mm with and aggressively sharpened in FocusMagic. Printed at 574 mm width in Lightroom.
- An A2 size print agressively sharpened in Lightroom and printed at 360 lp/mm from Lightroom
There is remarkably little difference between the two. Actually, I think Lightroom sharpening wins, but I may change my mind.
With the P45+, Lightroom has a tendency to create halos on edges. With the OLP filtered Sony's this seems not to be the case. Other raw converters like C1 and RawTherapee don't create halos but still produce plenty of artefacts. Personally, I feel that OLP less sensors are just a shame, they violate laws of sampling theory.
Best way to avoid artefacts is to decrease sensor pitch, but it doesn't work well with lenses with large beam angles. MFD needs OLP filtering badly but ignores that fact fully. May be MFD shooters care about detail, but don't care if that detail is fake or real…
The image below indicates how much aliasing that Planar lens produces on the P45+, top is Sony Alpha 77 with 16-80/3.5-4.5 zoom at 80/8 and bottom is P45+ with Planar 80/2.8 at f/8. Same distance, the idea is to demonstrate the effect of pixel pitch, not to compare different formats.
(http://echophoto.dnsalias.net/ekr/Articles/Aliasing2/Nyquist1_cropped.jpg)
Best regards
Erik
As I understand it , the capture sharpening is to get rid of the effects of the AA filter (which blurs thing slightly) so presumably if you don't have an AA filter, then there is no need for capture sharpening.
And output sharpening depends on the resolution used at print, and the type of paper.
keith
-
If you're camera is as sharp as you claim and is somewhat a recent model I'm assuming you're working with at least 16MP.
Yes, I've shot medium format on a tech cam for a while, so the older images are 33 megapixels and I'm changing to a 50 megapixel back now. So yes, you're right on a A4 sized print the resolution exceeds the printer capability. However, I'm interested in a workflow that scales up to 44" prints and beyond. At 44" I'm down at 100 ppi, and then I've found that one needs to be careful if the image should look "naturally fuzzy" up close, I'm pretty allergic to a look that even just hints pixels.
Yes, you need to nose the print to see it, but I just think it feels good to make a printed product that looks good at any distance, thus I'm interested in the upscaling and sharpening techniques.
-
Yes, I've shot medium format on a tech cam for a while, so the older images are 33 megapixels and I'm changing to a 50 megapixel back now. So yes, you're right on a A4 sized print the resolution exceeds the printer capability. However, I'm interested in a workflow that scales up to 44" prints and beyond. At 44" I'm down at 100 ppi, and then I've found that one needs to be careful if the image should look "naturally fuzzy" up close, I'm pretty allergic to a look that even just hints pixels.
Yes, you need to nose the print to see it, but I just think it feels good to make a printed product that looks good at any distance, thus I'm interested in the upscaling and sharpening techniques.
At that high capture resolution you have to consider how many pixels at native rez comprise each stair step square even if it's one to one. But at that tiny size you won't see it. Interpolation during uprezing to huge print sizes beyond 44" is going to give some softness to where you'll barely see the stair step artifact. I would guess higher radius after uprezing that high would have to be employed rather than using the Detail slider (ACR/LR) to add back micro contrast where alternatively at smaller rez would require adjusting Amount with a bit of Detail to get the same results.
You might try Nearest Neighbor resampling just to see what that does and how it sharpens up.
-
Maybe the combo of lens and back I use is immune to this, but I don't get an artifact that I can see unless I am zoomed in more than 120-150%
I love not having the AA filter softening effect. I'm so happy to get this ultra sharpness with MFDb, and one of the reasons I use a MF.
A few things...
When I print 24x36 from a 22mp MFDb I don't sharpen much at all, and they are tack sharp in print.
I also use C1 with Db files, and I don't sharpen to the point of seeing artifacts.
For workflow, you may consider Qimage as it sharpens and does a very good job so far in the week or so that I used it. It leaves the file alone, but you can also save an instance of the print job in a folder should you want to revisit with the parameters used.
Also, the lens used as mentioned has much to do with all this. This paired with the pixel pitch of the sensors. More current digital Apo Macro lens I use starts diffraction at F18 or 22. Many others much sooner.
I think you may also be crossing differences for intent of sharpening, as I remember I could slightly "over sharpen"(slight artifact) for offset printing as linescreens are at max 200, vs lf inkjet printers starting at 240 or so(I think HP and canon are around 260 vs Epson more in the 290 range?). This would result in a very sharp file. Small trade "secret" one picks up over the years, nothing you can't find online, but, for the newbies.....May also work for some canvas prints in some way, but I haven't tried.
So upresing I have also used BlowUp, or Perfect resize, and it does an OK, job, you have to set the right parameters for the right image, and it works good. Qimage has also a great upres feature and I stopped using Blowup so far.
I think what the software is usually doing is applying contrast to the "middle" tone areas in high contrast patterns, and when there is a large area of like color pixels it treats that area with less sharpening, etc...
Topaz Detail is sweet as you can define these areas manually to some degree, and you can squeeze a lot out of an image.....
But I am still puzzled how you can get stare casing out of the camera...
-
I don't think being without AA filter on the camera is a great idea, but in the MFD world that's the only option you have so that is what I have to get to work as well as possible.
I have noticed that I have a different taste on what looks good in terms of fine details, and might therefore want an unconventional workflow. When it comes to aliasing people in general doesn't seem to care, or rather the opposite if they don't see aliasing the lens is not sharp enough or there's too much diffraction. That's why D800E was more popular than D800, despite that I'd say the D800 (with AA filter) produces better looking fine details. Likewise upscalers like perfect resize is sharp but painterly looking. I want a smooth look that is as sharp as possible up close, but not showing any sign of that it's a digital picture consisting of pixels.
100% agree! I was starting to feel like a social misfit for having the same ideas, but now I'm feeling better; I'm not alone :)
-
Can someone provide and illustrate this concept?
On a photo forum soooo much can be explained via a simple image....>1000words
*(the picture above isn't doing much for me except tell me its low lit and soft, or not "the focus of interest in the image")
-
Can someone provide and illustrate this concept?
On a photo forum soooo much can be explained via a simple image....>1000words
*(the picture above isn't doing much for me except tell me its low lit and soft, or not "the focus of interest in the image")
I'll see what I can do, I'll do some experiments eventually and see if I can come up with a nice A/B comparison. After some initial test since my original post I'm not expecting any large difference, but possibly one can get a small edge on large prints. My printer is new though so there are like 1000 things to play with, I'm jumping like crazy between various things to test :-).
The crop with the bins was from a linear curve test shot from my balcony, unprocessed. It looks dull as it's unprocessed and no film curve,and subject is kind of monochrome. It was just an example of aliasing, but in that case I realize now that the false color aliasing is going to be a larger problem than staircasing.
If I shall guess I think it won't matter if I do a classic two-pass sharpening or one pass at the end, as Bart said the capture sharpening should not produce any new staircasing, it will just increase the contrast of what's already there. So I think the key will be how the upscaling and output sharpening is done.
I've looked around a bit and I don't think anyone has made the perfect upscaler yet, according to my taste. What I would like to have is a bicubic upscaler as base look, but with some special process just to eliminate staircasing. Ie like genuine fractals style concerning removing staircasing, but the soft and natural look of bicubic upscale.
-
For clarity of discussion, and my suggest...
Sharpen for intended use, not only for it to look good on your screen.
That is sometimes easier said than done trying to keep minimal copies of files. Let LR, or Qimage and otehr print-sharpen features take over. Often they do a good job.
-
Hi Anders,
Personally, I like the tools in Lightroom. My favourite settings are something like shown below.
I may open the image in Photoshop and apply some additional unsharp masking, say radius 2 and 15% amount. That increases contrast in low frequency details. Those details may be important in print.
With MFDB, I feel that Lightroom processing adds a halo that I cannot see with my Sony cameras, I would guess it compensates for the non existing OLP filter. For large prints I may use another raw converters, but in many cases I don't feel it matters. I hope that Adobe improves on demosaic in the next version of the processing pipe line.
That issue is discussed here: http://www.luminous-landscape.com/forum/index.php?topic=94812.0
Best regards
Erik
-
...
Personally, I like the tools in Lightroom. My favourite settings are something like shown below....
hai Erik, I use ACR and it looks the same..
on a d810 at base iso i never use masking ( =0)
Also i do not use any ( chromatic) noise reduction (i use only color noisereduction at high iso at 15 %)
Can you explain to me what is the benefit of both for you?
-
Hi,
Masking doesn't sharpen areas without edges. Sharpening always enhances noise. So, if you have a noise free image neither masking nor noise reduction are needed.
With the Alpha 99 I have right now and also with the P45+, I feel noise is much less an issue than before.
Luminance noise reduction reduces sharpness, color noise reduction much less. The P45+ has a lot of shadow noise, so I feel some color noise reduction is good to have. On the Alpha 99, I don't know…
Best regards
Erik
hai Erik, I use ACR and it looks the same..
on a d810 at base iso i never use masking ( =0)
Also i do not use any ( chromatic) noise reduction (i use only color noisereduction at high iso at 15 %)
Can you explain to me what is the benefit of both for you?
-
Also, the lens used as mentioned has much to do with all this. This paired with the pixel pitch of the sensors. More current digital Apo Macro lens I use starts diffraction at F18 or 22. Many others much sooner.
Diffraction limits the resolution at all apertures, but when we say that a lens is diffraction limited at a certain f/stop, this means that its resolution is limited only by diffraction rather than aberrations, beginning at that aperture. A lens that is diffraction limited at f/18 or f/22 is not a well performing lens as shown by the limiting resolutions shown in the table below. A MTF50 of 35 lp/mm is not impressive for a dSLR or MFDB, but it would be OK with an 8 x 10 inch view camera. With such a lens, aberrations predominate at larger apertures and the potential loss of resolution due to stopping down is not that apparent until the aperture is small enough to eliminate the predominant effects of aberrations. With an excellent lens, loss of resolution from stopping down is apparent not far from the maximum aperture.
(http://bjanes.smugmug.com/Photography/Aperture-Series/i-P8ZwzT6/0/O/ResolutionLimits.png)
The effect of diffraction is independent of pixel pitch, and for a given aperture, there is no loss of system resolution with a finer pixel pitch. However, if you want to make use of the increased resolution offered by a fine pixel pitch you shouldn't stop down so that the diffraction spot (Airy disc) is more than approximately twice the pixel size (assuming a Bayer array sensor). These concepts are well reviewed here (http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm).
Regards,
Bill
-
It sounds like I have to read this a couple times and the link provided. Thank you for them.....
But, I'm not sure YET, what this means in usage. Maybe you are correcting my terminology in using deffraction vs aberration? Here is my observation....
I use both a current macro digi Schneider and Rod lenses with use on a 11mircon pixel on 4x5 and with a loop I can see the focus get fuzzy beyond ~F22.
I was under the impression that this is deffraction caused by reflective circles of confusion due to the blades shape of the shutters? I maybe off, but its what I remember when working and reading up on this and testing things out. My memory works less and less like digital device every year :-)
-
It sounds like I have to read this a couple times and the link provided. Thank you for them.....
But, I'm not sure YET, what this means in usage. Maybe you are correcting my terminology in using deffraction vs aberration? Here is my observation....
I use both a current macro digi Schneider and Rod lenses with use on a 11mircon pixel on 4x5 and with a loop I can see the focus get fuzzy beyond ~F22.
I was under the impression that this is deffraction caused by reflective circles of confusion due to the blades shape of the shutters? I maybe off, but its what I remember when working and reading up on this and testing things out. My memory works less and less like digital device every year :-)
Hi Phil,
At f/22, the diffraction pattern diameter for Green light (which is the dominant contributor to Luminance resolution) of e.g. 555 nanometer wavelength, is 2.44 x 0.555 x 22 = 29.8 micron. That means that the finest detail is blurred by diffraction alone to a size that's more than 2 (11 imicron pitch) pixels in diameter. Added to that is the blur that is caused by residual lens aberrations. So I'm not surprised that you could see that with a loupe, it's as expected from the laws of physics.
You'll be able and see the image get fussy at wider apertures already if you use smaller pitch sensels. Smaller pitch sensels can pick up deterioration from smaller diffraction blur diameters more accurately, and thus at wider apertures.
Cheers,
Bart
-
It sounds like I have to read this a couple times and the link provided. Thank you for them.....
But, I'm not sure YET, what this means in usage. Maybe you are correcting my terminology in using deffraction vs aberration? Here is my observation....
I use both a current macro digi Schneider and Rod lenses with use on a 11mircon pixel on 4x5 and with a loop I can see the focus get fuzzy beyond ~F22.
I was under the impression that this is deffraction caused by reflective circles of confusion due to the blades shape of the shutters? I maybe off, but its what I remember when working and reading up on this and testing things out. My memory works less and less like digital device every year :-
Phil,
I see that Bart has already commented regarding the effects of the size of the Airy disc with respect to 11 micron pixels. I am confused as to what system you are using. You mention 4x5 (?inch) and 11 um pixels, but there are no 4x5 inch digital sensors. 35 lp/mm resolution looks much better with 4x5 inch than 24x36 mm, since the latter image needs more magnification for a given print size.
Regards,
Bill
-
The explanation sounds factual Bart, yet perhaps the results would be backpedaling with smaller pixels...Or I'm misunderstanding?
We are trying to achieve more sharpness/more dof,
Smaller pitch sensels can pick up deterioration from smaller diffraction blur diameters more accurately, and thus at wider apertures.
So should I gather from this that for someone looking to stop down more/high F without deffraction in order to get more DoF, smaller pixels is NOT the answer, but on the contrary.
Yes Bill, the 4x5 is regarding the "tunnel". Somewhere I mentioned its a 22mpixel MFDb.
there is no loss of system resolution with a finer pixel pitch
With smaller pitch is the DoF increased or decreased? I think at this point another factor plays in the mix, and that is resolution. Take a finely polished 35cm marble for example, and try to fill the frame of a 22mp Db. Then take a 40mp Db on same setup, and try filling that frame...By physics, would they have the same DoF? But the 40mp having the 18mp added size in output?
(Can you tell I have yet to read that link :-)
-
Hi,
If you look at an enlargement of given size, DoF is not dependent on pixel size, nor is diffraction.
Smaller pixels will allow better sharpening, so it may be possible to compensate for diffraction better with smaller pixels.
On the other hand, it is conceiveable to print larger with smaller pixels, and with a larger print size the CoC on sensor needs to be smaller. Larger prints are more demanding.
Simply enough, smaller pixels are benefitial, except that DR will be slightly better with larger pixels and the raw files will be larger with smaller pixels.
Best regards
Erik
The explanation sounds factual Bart, yet perhaps the results would be backpedaling with smaller pixels...Or I'm misunderstanding?
We are trying to achieve more sharpness/more dof,
So should I gather from this that for someone looking to stop down more/high F without deffraction in order to get more DoF, smaller pixels is NOT the answer, but on the contrary.
Yes Bill, the 4x5 is regarding the "tunnel". Somewhere I mentioned its a 22mpixel MFDb.
With smaller pitch is the DoF increased or decreased? I think at this point another factor plays in the mix, and that is resolution. Take a finely polished 35cm marble for example, and try to fill the frame of a 22mp Db. Then take a 40mp Db on same setup, and try filling that frame...By physics, would they have the same DoF? But the 40mp having the 18mp added size in output?
(Can you tell I have yet to read that link :-)
-
that helps. So I understand Bart and Bill in this, just that they were not what I was hoping to hear :-)
I guess your saying this makes sense...that DoF nor defraction, is effected by pixel size. Once you factor in distance and focal I think it makes a difference....
As in.... If you have a subject at X distance, and use a zoom to get frame fill of the subject with a smaller pixel sensor(yet high resolution), vs that same subject with less zoom, but camera closer to subject to fill frame with larger pixel sensor(somehwat lower resolution), I'm pretty surer the high resolution with smaller pixels will have more DoF. I think that was my findings when testing this.
So I take it the sensor makers got all the issues with overlay lenses resolved with the color bleeding issues and such.
So the new crop of sensors prove to do a better job than the 22mp 11micron chips from years ago....I think we are down to 4-6micron chips these days, right? At least in PhaseOne backs and maybe comparable to Sony 36mp pixel size?
I hope this info is helping the OP. It is helping me clarify things that have started to get foggy, and now clear:-)