Epson V850 scan with 100% crop

[Steven Ruttenberg]

Color Cast Removed, Chamonix 45HS-1, Nikkor 75mm f/4.5 at f/32, front tilt, fall. Extar 100. Wet scanned using Betterscanning adjustable holder, ANR glass, Kami Fluid and Optical Mylar. Epson V850 at 4200dpi. I downsized from 4200dpi to 300dpi and saved as jpg, sRGB. I also included the 100% crop of the center of the frame. This was about 2/3 of the way into the image from where I was standing. I still need to fine tune the focus adjustment of the V850 using the holder, but this is pretty close. I don't think it did too bad. Unfortunately, flickr keeps over-saturating uploaded images.

https://farm5.staticflickr.com/4877/44861951945_95d5ed1335_b.jpg (https://flic.kr/p/2bmiem6)20181027_0129_Working_201891105 (https://flic.kr/p/2bmiem6) by Steven Ruttenberg (https://www.flickr.com/photos/157376714@N08/), on Flickr


https://farm5.staticflickr.com/4910/30834673617_f9338f19a1_o.jpg (https://flic.kr/p/NYKLcK)Center Crop of 20181027_0129_Working_201891105 (https://flic.kr/p/NYKLcK) by Steven Ruttenberg (https://www.flickr.com/photos/157376714@N08/), on Flickr

[Peter De Smidt]

Way better!

[Steven Ruttenberg]

Thank you!

[Pere Casals]

Steven,

V850 scans come less digitally optimized than those from Pro scanners, those have amazing (in firmware or in driver) sharpening optimization that increases the effect until it detects overshot condition.

With a V850 you don't have that automation and always have to adjust sharpening in Ps, these are the settings I would use for this image:

184255

(sharpening 1.4pix , 80%)

This well known: https://petapixel.com/2017/05/01/16000-photo-scanner-vs-500-scanner/

From that Petapixel test, we have to realize that as the scan width is increased the V850 shines the more, because the V850 mantains the same resolving power for MF than for LF, while the other zooms out until 2000 samples per inch.

Another advice is scanning 16bits and not using the limits suggested by the scanning software that tend to clip in the ends of dynamic range to deliver a good tonal curve for the mids.

So a bit it's like developing a RAW DSLR image, you feed photoshop with a dull image that has "more scene dynamic range" than the presentation medium has, and then you have to bend curves, clip and locally edit to obtain what you want.

IMHO this scan is not well done because you have little detail in the saturated yellows, you may post an screenshot of the scanning settings.

You should obtain at least a color gradation like this one, it's a V750 crop of a 35mm shot:

https://c1.staticflickr.com/5/4888/45728063582_8a742bb0cd_b.jpg

[Tin Can]

Since we know Steven is setting up an enlarger, perhaps a comparison of similar scans to optical wet print enlargement.

Do we have access to any examples?



Steven,



This well known: https://petapixel.com/2017/05/01/16000-photo-scanner-vs-500-scanner/

[mdarnton]

I observe that the grain in the expensive scanner detail in the Petapixel article is coarser than the Epson detail, so I took the side-by-side example into Photoshop and sharpened the Epson side until the grain was equal. At that point, to me, the Epson side actually appears noticeably sharper.

[Ted Baker]

Unfortunately, flickr keeps over-saturating uploaded images.


There are few possible reasons for this, if you are happy to make available the exact jpg that you uploaded to flickr, I would be happy to make an analysis, when I get time. Doesn't have to be this image just one that has the problem as described.

[Pere Casals]

Since we know Steven is setting up an enlarger, perhaps a comparison of similar scans to optical wet print enlargement.

Do we have access to any examples?

I'd prepare some microscope images...

[Steven Ruttenberg]

On the sharpening use high pass and try to avoid too much of it. I rarely use another method, but open to trying. I find too much sharpening or wrong kind while age may look sharper it ends up being garish.

[Steven Ruttenberg]

There are few possible reasons for this, if you are happy to make available the exact jpg that you uploaded to flickr, I would be happy to make an analysis, when I get time. Doesn't have to be this image just one that has the problem as described.

Sure. I will put in dropbox and send link. Maybe I am just seeing things.

[Steven Ruttenberg]

Here is another shot, If I have time tonight, I will make a 100% crop, but this was taken with Tmax100 at night 30 minute exposure (15 minutes measured) with 210mm G-Claron lens at f/11. I scanned in same as fall colors image and sharpened the same. One thing to note on fall colors, is the distortion from ultra wide angle (75mm) and there was some wind. I think if you nail focus when taking the picture, how it scans is related more to how you develop image and the actual scene/exposure. Get any of those to far off and no amount of scanning technique can save the image (maybe make it useable though)

https://farm2.staticflickr.com/1741/42617080442_1c116370d8_o.jpg (https://flic.kr/p/27VVFg1)20180519_0046_20180607_Working_Flattened (https://flic.kr/p/27VVFg1) by Steven Ruttenberg (https://www.flickr.com/photos/157376714@N08/), on Flickr

[Pere Casals]

For the sharpening here you have a good guide:

https://www.cambridgeincolour.com/tutorials/image-sharpening.htm

You should always use adavanced algorithms for that, this makes a great difference ! USM is nice leap forward.


________

This the kind of algorithms may be used inside Pro gear to adjust sharpening automaticly to the best point: http://mepro.bjtu.edu.cn/res/papers/2011/Unsharp%20Masking%20Sharpening%20Detection%20via%20Overshot%20Artifacts%20Analysis.pdf

That paper describes algorithms for detecting if an image was sharpened, but the same also serves to find the sweet point for automated in firmware procedures.

To me sharpening it's a complex issue, and proficiency with it it's critical to obtain a sound result in the digital workflow.

[Steven Ruttenberg]

Thanks for the info. I have steered away from ism after using it for years. I have been leaning towards sharpening that is subtle and almost unnoticeable. I find images that are perfectly sharpened to be hard on on the eyes. Much like,oder 4K tvs and movies. While they are undoubtedly technically perfect they appear fake and fabricated. No character, very anticeptic. Much like modern camera lenses. Technically perfect, but very sterile image reproduction.

I am sure though there are elements of the references that I can use on images. Although, the second article has some calculus and linear algebra. Haven't done that in tears. I may try some sharpening at scan time and see what happens.

[Ken Lee]

With respect to sharpening, you might find this brief article helpful: Avoid Sharpening Artifacts: Sharpen the Mid-Tones Only (http://www.kennethleegallery.com/html/tech/index.php#Sharpening).

It's written for Photoshop, where there are often multiple ways to accomplish the same goal.

[Pere Casals]

Ken, I like your approach, good advice, but I'd add that in the layers mix, instead a not sharpened image we may use one with a milder sharpening, because shadows and highlights may also benefit from some milder sharpening.

Now Ps CC has a Smart Sharpen filter that would help because of not destructive feature.

In a perfect world this would no be necessary if Ps had nice ASOC type algorithms (Adaptive Sharpening with Overshoot Control), or ADSOC for the directional version, or "Unsharp Masking Sharpening Detection via Overshoot Artifacts Analysis" to locally adjust the effect.


IMHO it also has to be mentioned that in a similar way (with masks) we should do a local edition. In a portrait we may want to sharpen well eyes and not the cheek. If we portray wrinkles in the skin 0.001mm deeper than they are then we can be killed :)

A bit off topic... illumination reflections in the eyes are usually out of focus, as optically they are farther than the eyes, in a headshot it may have a remarkable effect to custom sharpen it.

hmmm, digital is amazing for those things, but I find that the atmosphere obtained by (well used!) dedicated classic glasses cannot be surpassed by any kind of digital edition. I would like to understand why...


______________

http://www.johnpaulcaponigro.com/blog/7747/7-sharpening-artifacts-to-avoid/
https://www.youtube.com/watch?v=0P8IQ2SfRCE
https://www.youtube.com/watch?v=edpPvq8OUYo
https://www.youtube.com/watch?v=e3jToF99bpw


Also it's worth to explore Nik Sharpener Pro plugin for Ps (etc) , IIRC Nik is now owned by DoX
https://www.youtube.com/watch?v=y-_QV-LFoQk

[Peter De Smidt]

https://photographylife.com/landscapes/advanced-post-processing-tips-three-step-sharpening

Like Steven, I find that restraint is best when it comes to sharpening. With scans, I do use deconvolution capture sharpening, as per the article, right at the start. It makes grains look like grains, and it help with spotting. In Photoshop this is via the camera raw filter. I haven't been doing any local sharpening lately. For output sharpening, I often totally mask it off from smooth areas.

[Ken Lee]

Ken, I like your approach, good advice, but I'd add that in the layers mix, instead a not sharpened image we may use one with a milder sharpening, because shadows and highlights may also benefit from some milder sharpening.

Yes, it's even better to sharpen different sections of the tonal scale independently.

Previously I wrote something called Sharpen the Dark, Mid and High Values Separately but I thought it was too much work for the average person, so I replaced it with the current article.

[Pere Casals]

Yes, it's even better to sharpen different sections of the tonal scale independently.

Previously I wrote something called Sharpen the Dark, Mid and High Values Separately but I thought it was too much work for the average person, so I replaced it with the current article.

Yes, some times the easiest way it's the best...

My perception is that sharpening tools are being smarter every day, perhaps in the near future it would be easier to make a perfect job.

[Ken Lee]

Yes, some times the easiest way it's the best...

My perception is that sharpening tools are being smarter every day, perhaps in the near future it would be easier to make a perfect job.

These tools are probably aimed at professionals who need to sharpen a lot of photographs in quick time at low cost.

For example, let's say I shoot 2,000 photos of a wedding. Only 5 images will be printed on paper and the rest of the images will appear online only, for people to view briefly on their smart phones while waiting for the train. I would gladly pay for a tool which will let me "sharpen" them all in the click of a button and automatically load them to a wedding web site. If the tool can automatically make the colors "vibrant" at the same time, that's even better.

Perhaps in the not-so-distant future people will hire a robot to take the wedding photos, not just process them.

After that the minister will be a robot (see https://www.engadget.com/2018/11/08/chinese-news-agency-ai-anchors/).

Eventually, you will marry a robot or hire 2 robots to get married for you... then the photographs will be perfect !

[Pere Casals]

These tools are probably aimed at professionals who need to sharpen a lot of photographs in quick time at low cost.


Image enhancing software started being a competitive arena in the digital minilab era, all manufacturers claimed to have better abilities than the others. Today that software recognizes the scene... if a portrait it is guessed male/female, race, age, lifestyle...

I guess that Ps (and the like) will easily evolve the algorithms to work locally in a smarter fashion. Beyond locally limiting the effect by monitoring ovreshot, defocus areas can be located to use agressive denoise there, while in the areas with microcontrast the strategy would be another. Well, just an slider for more/less and making the software think for us.

One thing I do sometimes to make a mask for defocus areas is an edge detection followed by an ample dilation to fill the holes, and then an ample erosion to restore the outer contour, with this simple way we can selectively mask OOF/DOF areas to denoise or to sharpen where it has to be done, Ps developers may be intensively working in reliable automation for those kinds of things...

[Steven Ruttenberg]

Here is where I would hesitate. Anything that is automatic, semi-automatic with little or no control over the process. Ie, there are processes that are automatic and allow you to change certain parameters within the process, but never give the end use complete control over what happens. The problem with even a wedding and 2,000 pictures is that maybe 70% of them are all within a certain range if you will where you can apply a canned sharpening routine, even a so-called smart sharpen routine, and they look okay to look at on a cell phone. The other 30% fall outside the range and will look horrible because it is that 30% that will be noticed. When it is the whole lot will be considered bad.

I want tools that I can automate myself, yet have 1000% control over every aspect of the process. I tried photoshop smart sharpen and didn't like it. I used to use USM, but it was a bear to use most of the time. I like high pass for now as it seems to allow me to do a good job overall and I can use mid tone sharpening with it. But even it has its short comings. I tried Nik pre,post raw sharpening, and wasn't impressed in the least. Perhaps it is best to sharpen at capture (when scanning) and avoid sharpening for output. Just a thought. I know my negatives appear to be much sharper than any scan I have made.

Anyway, I am gonna do some playing around this weekend if I don't over eat since it is Veterans Day/weekend and we get all kinds of free meals the whole weekend. :)