Low-Resolution Shooting and Post-Processing

[xpatUSA]

Being one committed to monitor viewing with no printing involved, the need for resolution is less important for my good self. So, with my 1920x1200px monitor, my 2268x1512px Sigma SD10 is good enough.

But, for most here, such an insanely low resolution comes at a price. That is to say, detail in the image that is close to the imager resolution of 9.12um, will inevitably lack local contrast.

So it is that, in this post, I would like to show that under such circumstances all is is not lost!

I present Sigma's conversion of a shot with all adjustments neutral insofar as local contrast is concerned:



To me, even though the focus is accurate enough, the acutance of the rendered image is quite low, what do you think?

So I went into RawTherapee and did quite a bit of wavelet stuff and a bit of RL de-convolution sharpening and more:



I hope y'all can see the difference.

For those familiar with RawTherapee, here's the profile:

http://kronometric.org/phot/post/CiC...d-sunlight.pp3

Comments welcome!

[pnodrog]

Second version definitely looks better in terms of sharpness etc. I was about to comment that it still seemed to be lacking a bit in overall contrast but just rembered my monitor is very misleading at this time of day with the curtains open behind me.

[Shadowman]

At first glance I liked the first image as the colors were more vibrant, I decided to use Lightbox to get a quick change and to me it is confirmed that the first is much more desirable at least to my eyes.

[Mike Buckley]

For me, the first one is aesthetically more pleasing and the second one is technically more pleasing. Given the option of having to choose one over the other when as in this case neither is hugely lacking, I will always go with the aesthetically more pleasing image.

[Manfred M]

What I notice more than the sharpness is the colour shift between the two images. The first one is a touch warmer and the second one seems to be more neutral colour wise.

At normal viewing distances I tend not to get hung up on sharpness, but the overall look just. I view images on my computer screen in a way where I see the whole image and don't bother pixel peeping. I suspect that the second image is probably more accurate when it comes to colours, but is colour accuracy what you are after?

[dje]

Yes Ted the second image is certainly sharper, you can see it in the grass especially. However as Manfred says, there seems to be a color shift from one image to the other, almost like the White Balance was changed (made a little cooler)?

Dave

[Rent]

First image to me is definitely better, the second looks washed out. If one didn't see the first, the second would probably be OK
Roy

[xpatUSA]

Thanks to all for your comments.

A pity that the thread has veered off into comparisons of global contrast and color when the original post was more about local contrast (sharpness). To get back on track, please just compare the satellite dish, the outdoor fireplace, the solar panel and nearby grass.

Perhaps I should re-do the second image to restore the color and apply more global contrast?

[Manfred M]

Thanks to all for your comments.

A pity that the thread has veered off into comparisons of global contrast and color when the original post was more about local contrast (sharpness). To get back on track, please just compare the satellite dish, the outdoor fireplace, the solar panel and nearby grass.

Perhaps I should re-do the second image to restore the color and apply more global contrast?

I think the comments were valuable in establishing that local contrast is just one element in the overall image that the viewer needs to assess. Other processing "concerns" can be given greater weight by the viewer than microcontrast.

If I were to judge the three images you posted in a side-by-side fashion; the first one needs more micro-contrast. the second one looks fine and the right-most one looks oversharpened to me.

[Mike Buckley]

Agreed with all points made by Manfred in his most recent post.

[xpatUSA]

Perhaps I should clarify the purpose of this thread.

Because I do not print and my 1920x1200px monitor satisfies my viewing needs, I do not need a high-resolution camera. But the use of a 2268x1512px camera brings a penalty of low LOCAL contrast, aka micro-contrast for detail that is near Nyquist in the image plane. Such lack of contrast appears as a lack of acutance in the converted image, even with a good, well-focused lens. Nyquist for the camera is 55 lp/mm in the image plane, BTW, quite within the reach of a decent lens. Also the sensor has no AA blur filter on it.

Would like to hear comments on the above, if any (other than 'buy a 100MP Canon' of course ).

Then I claimed that post-processing is needed to improve such images and provided illustrations to show that such improvement is possible. The illustrations were intended to show improvements in acutance, or micro-detail or local-contrast. The illustrations were not intended to be pleasing masterpieces and were not posted for C&C.

If my illustrations have failed to show improvements in acutance, or micro-detail, or local-contrast, or whatever you call it, I do apologize.

[Mike Buckley]

If my illustrations have failed to show improvements in acutance, or micro-detail, or local-contrast, or whatever you call it, I do apologize.

There is certainly no need to apologize.

My hunch is that not everyone will use the same criteria when determining whether there are improvements in acutance and the like. As an example, I think the third image displays too much acutance, which for me means there is no improvement. Similarly, if I add too much sugar to my ice tea, it definitely gets sweeter though that doesn't mean there has been an improvement in the sweetness.

Based on all of that, I wonder if it would be helpful for you to ask simply if there has been an increase rather than an improvement in acutance. Indeed, the second and third images for me display progressively more increase but only the second one displays an improvement.

[Manfred M]

Ted - I think we did understand the question, but part of our response was that from a "global" viewpoint, micro contrast is just one of the ingredients the photographer needs to consider in creating a viable image

That being said, there needs to be an appropriate level of micro contrast applied. When looking at the three images you posted together, this clearly shows that a side-by-side comparison (which helps emphasize your point). All three of the images would probably stand a cursory inspection and the amount of sharpening might or might not be picked up by the casual viewer.

However, when all three images are shown side by side; the left-most image needs more micro-contrast. The middle one looks good and the rightmost one has too much micro-contrast.

[xpatUSA]

Yes, gentlemen, the third image was well over-the-top as y'all rightly say.

However, it would never have existed had there not been comments on the color and on the global contrast, neither of which were the subject of the OP. A classical example of topic drift, in fact.

So, in terms of the wider aspects of general image quality, pleasing color and the meaning of "improvement", I bow to the recent views expressed.

[Manfred M]

However, it would never have existed had there not been comments on the color and on the global contrast, neither of which were the subject of the OP. A classical example of topic drift, in fact.

Perhaps Ted, but I would suggest that fixing all variables other than micro contrast would be the only way to do an accurate assessment of the impact on the image. As this wasn't done, it impacted the viewer's assessment of the images as several variables came into play.

[Mike Buckley]

A classical example of topic drift, in fact.

No, not in fact; only in your and perhaps others' opinion. You wrote in your first post that "all is not lost" and then made numerous changes in your image. It did not occur to me that I was expected to ignore all those other changes and that when I didn't, topic drift would be attributed at least in part to my posts.

[xpatUSA]

OK, gentlemen, I fold. I broke my own rule: when comparing this with that, keep all other variables the same.

[dje]

Because I do not print and my 1920x1200px monitor satisfies my viewing needs, I do not need a high-resolution camera. But the use of a 2268x1512px camera brings a penalty of low LOCAL contrast, aka micro-contrast for detail that is near Nyquist in the image plane. Such lack of contrast appears as a lack of acutance in the converted image, even with a good, well-focused lens. Nyquist for the camera is 55 lp/mm in the image plane, BTW, quite within the reach of a decent lens. Also the sensor has no AA blur filter on it.

Hi Ted

The AA filter is usually the biggest contributor to lack of sharpness in a low resolution camera as it's main purpose is to reduce the MTF of the system to a small value around and above the Nyquist frequency, in order to minimise moire (as I'm sure you know). So if the Nyquist frequency is low, the MTF will be low at relatively low frequencies. You have an advantage in the sharpness race with your camera as it has no AA. Nevertheless, sharpening can make a significant difference to image quality, even without pixel peeping. Of course with a camera resolution of 2268x1512 px you are virtually pixel peeping all the time when viewing the image on a 1920 x 1200 px monitor!

I remember revisiting some images from my first digital camera which was a 2MP Canon. These jpegs would have had in-camera sharpening but judicious use of some de-convolution sharpening (Topaz Detail) really made them pop.

As an aside, I interpret the term "local contrast" differently to you. I don't see it as acutance but rather as an emphasis on contrast variation over areas covering maybe a couple of hundred pixels. In other words something in between micro contrast and global contrast.

Dave

[xpatUSA]

Hi Ted

The AA filter is usually the biggest contributor to lack of sharpness in a low resolution camera as it's main purpose is to reduce the MTF of the system to a small value around and above the Nyquist frequency, in order to minimise moire (as I'm sure you know). So if the Nyquist frequency is low, the MTF will be low at relatively low frequencies. You have an advantage in the sharpness race with your camera as it has no AA. Nevertheless, sharpening can make a significant difference to image quality, even without pixel peeping. Of course with a camera resolution of 2268x1512 px you are virtually pixel peeping all the time when viewing the image on a 1920 x 1200 px monitor!

Hi Dave, thanks for the input.

I remember revisiting some images from my first digital camera which was a 2MP Canon. These jpegs would have had in-camera sharpening but judicious use of some de-convolution sharpening (Topaz Detail) really made them pop.

Indeed, I too like the idea of de-convolution.

As an aside, I interpret the term "local contrast" differently to you. I don't see it as acutance but rather as an emphasis on contrast variation over areas covering maybe a couple of hundred pixels. In other words something in between micro contrast and global contrast.

Yes, I used the term loosely as is de rigeur in these fora. I was really talking about micro-contrast. As to how many pixels that covers, that is probably better expressed here, sorta:

http://rawpedia.rawtherapee.com/Wavelets#Total_Levels
.

[dje]

Hi Dave, thanks for the input.



Indeed, I too like the idea of de-convolution.



Yes, I used the term loosely as is de rigeur in these fora. I was really talking about micro-contrast. As to how many pixels that covers, that is probably better expressed here, sorta:

http://rawpedia.rawtherapee.com/Wavelets#Total_Levels
.

Interesting article Ted. I've never really looked into wavelet sharpening but I probably should! It looks like you need a PhD to use it in Rawtherapee though

Dave