Railfan Technology > Photoshop Re-size, Smart Sharpen, and Output Methods

I'm qualifying digital photography as railfan technology, thus I wanted to play around here with various methods for outputting images from Photoshop for upload on TO.

I find that the 1000-pixel max width files are a challenge for me working with a Retina Display, as the resulting images usually look over-sharpened or over-pixelated when exported and viewed as scaled jpgs in Preview or after upload. (MacOS rendering upscales images in Safari and Preview, whereas I can view at 1:1 in Lightroom and Photoshop.) I'm shooting either FF Nikon or micro 4/3s cameras with moderate to high resolution sensors and good glass. The results before resize or after moderate resize look fantastic on my Mac, but I'm struggling to get something I like for posting here at 1000-pxels max width and 500kb max file size.

I've been playing around with various techniques related to when I resize images, using Bicubic versus Bicubic Sharper for image size reduction, when and how much sharpening I apply using the Smart Sharpen filter in PS, using the Export As versus the legacy Save for Web outputs, and finally, I wanted to see what the website will do when I upload images wider than 1000-pixels wide (assuming it will auto-scale them to 1000-pixels).

My typical workflow is as follows:
Import Raw into Adobe Lightroom, make adjustments/process to taste, open image in Photoshop, covert image to a Smart Object, resize the image, apply Smart Sharpen, add my signature as a layer pasted from a .psd, and then finally export the final file.

So, for my first test, attached are images that are 1) as published on the Western Board earlier today which was exported as a 1000-pixel wide jpg, 2) a 1500-pixel wide copy of the image saved at a lower quality to keep the file size below 500kb, and 3) a 2000-pixel wide image at a very low quality, again to stay below 500kb.

If it's a problem to have multiple copies of the same file, I'll pull them down. Just wanted to see a comparison to determine if there is/are better options to get the best quality uploaded.

Thanks,
Sean

---

---

---

If you look at the image info all three are at 1000 pixels, the TO compression software has been the topic of a few discussion. The common way to 'game the system' is to resize the image to 1000 or 999 pixels on your own workstation before attaching it. This appears to keep the compression software away from the image, so it doesn't add noise or color adjustments, or so we think.

Regarding PSE or Lightroom 'workflows', it's popular to have your own workflow. A number of phottogers process in one s/w then move it to another, or even back and forth. I have noted that some PSE actions work better with smaller files, but others work better with large files. Sharpen and the Auto functions seem to be most sensitive to file size. Then there's also compression noise, etc. Compression noise can also be described as the result of removing pixels to make the file smaller.

To make any sense out of it I went to Landscape Photographers or wildlife Photographers for examples, since there is some commonality between the fields. Many of them have YT presentations and many have pay for classes, some just like to show off. Here's a link to one photographer that sells his info but also has guided tours. In his freebie demos he has some ideas about 'workflow' that seems to make sense. Some pretty sensible equipment discussion in there as well:

https://backcountrygallery.com/

https://backcountrygallery.com/new-video-workshop-noise-reduction-using-lr-ps/



Edited 1 time(s). Last edit at 05/26/20 21:39 by TCnR.

I seem to have better (... i.e. "sharper") TO display results when I post scanned slide photos at 1500 pixels and let the TO software "fix" it.  To my eyes, the images turn out visibly sharper for some reason.  But, I agree that it is often time-consuming and frustrating to get just the right "mix" for a TO presentation.

On the other hand, I have very few issues with TO's presentation when I post images taken with a digital camera and I think that also applies to a lot of the photos posted by others.

Definitely a "still learning" process for me.

Greg Stadter
Phoenix, AZ

Thanks TCnR. I've followed some of the other threads on this topic, and I'm probably beating a dead horse. I was mainly curious to test if there was any discernible difference in the final image quality of the three scenarios. I think much of my issue is Mac Retina Display related. My work-flow in both Adobe Lightroom and Photoshop is solid. No issues in any other other situations, from making 13"x19" prints on my R2880, posting high resolution images for other sites, or down to posting on Instagram at 1080 pixels wide. (Note that in the case of Instagram, while images are only 1080 pixels wide and are generally being viewed on a smaller device display, and hence they look sharp, they also display at a true 1:1 image pixel to display pixel ratio when viewed on a laptop or desktop browser, so they continue to look appropriately sharp.)

My problem seems to relate to the final output at 1000 pixels max width, and subject size in the resulting image. There is a lot of fine detail that is lost in downsizing to 1000 pixels and then viewing on a Retina display, as both the preview tool and web browsers will dynamically scale up the images in MacOS, which tends to make them somewhat pixelated looking. Websites that use responsive design or adaptive design are browser aware, and they will serve up higher resolution images for Retina displays. That doesn't exactly work for this site, given bandwidth constraints and member content sharing.

If I have the same 1000 pixel image open in Photoshop, I can view it at a true one-to-one image pixel to display pixel ratio, which makes it look much sharper but also about half the width compared to what I see in Preview or in a browser like Safari. (1:1 in Photoshop for a 1000 pixel wide image is around 4.5" on my MacBook Pro, while the same image in Safari displays around 8.5" wide.) It's interesting to note that in my experiment above, the resultant images all look more or less the same. Therefore, a higher quality 1000 pixel wide image will look the same on the site as a medium quality 1500 pixel wide image downsized after upload or a very low quality 2000 pixel wide image downsized after upload. This makes sense to a degree, as all images had around 470KB of information. I tend to find on the larger format Retina Display devices (iPad, MacBook Pro, iMac), once the image size drops below ~1200-1500 pixels wide, it doesn't tend to render well when dynamically resized for web browsers, etc.

Sean

TCnR Wrote:
-------------------------------------------------------
> If you look at the image info all three are at
> 1000 pixels, the TO compression software has been
> the topic of a few discussion. The common way to
> 'game the system' is to resize the image to 1000
> or 999 pixels on your own workstation before
> attaching it. This appears to keep the compression
> software away from the image, so it doesn't add
> noise or color adjustments, or so we think.
>
> Regarding PSE or Lightroom 'workflows', it's
> popular to have your own workflow. A number of
> phottogers process in one s/w then move it to
> another, or even back and forth. I have noted that
> some PSE actions work better with smaller files,
> but others work better with large files. Sharpen
> and the Auto functions seem to be most sensitive
> to file size. Then there's also compression noise,
> etc. Compression noise can also be described as
> the result of removing pixels to make the file
> smaller.
>
> To make any sense out of it I went to Landscape
> Photographers or wildlife Photographers for
> examples, since there is some commonality between
> the fields. Many of them have YT presentations and
> many have pay for classes, some just like to show
> off. Here's a link to one photographer that sells
> his info but also has guided tours. In his freebie
> demos he has some ideas about 'workflow' that
> seems to make sense. Some pretty sensible
> equipment discussion in there as well:
>
> https://backcountrygallery.com/
>
> https://backcountrygallery.com/new-video-workshop-
> noise-reduction-using-lr-ps/

I'm sensing a difference in the three photos, more in the background trees, the shadows and the greens are different. Which is pretty minor, I do see great resolution in all three, nice work.

Using a Dell U2312HM and firefox.

Good to hear some ideas on the general subject. There's also great work on Railpictures dot Net, if you can deal with the rules and guidelines.

I'm probably making a mountain out of a mole hill here. Do the images above look appropriately sharp (not too much, not too little) and detailed to you? Compared to others posted on the site? Any feedback would be appreciated.

It seems to me that digitally captured work shared here from some others, but not all, looks a little smoother while still being sharp and detailed in terms of transitions in areas of edge contrast. It might just be that I'm looking at the original raw images from 24MP FF and 20MP micro-4/3rds cameras, which have 6 or 5 times the linear resolution of the 1000 pixel wide images I'm sharing here. So perhaps I've just been "spoiled" by seeing what the raw images have to offer.

Thoughts?

Thanks,
Sean

E25 Wrote:
-------------------------------------------------------
> I seem to have better (... i.e. "sharper") TO
> display results when I post scanned slide photos
> at 1500 pixels and let the TO software "fix" it. 
> To my eyes, the images turn out visibly sharper
> for some reason.  But, I agree that it is often
> time-consuming and frustrating to get just the
> right "mix" for a TO presentation.
>
> On the other hand, I have very few issues with
> TO's presentation when I post images taken with a
> digital camera and I think that also applies to a
> lot of the photos posted by others.
>
> Definitely a "still learning" process for me.

Thanks, that's helpful and good to know.

I am familiar with that other railroad photo sharing website and their rules and regulations. I have a gallery there, though I haven't gotten around to posting anything there for over a decade. I haven't been trackside much until recently when we got a place up near Donner Summit last fall, mostly because I have 10 and 12 year old kids. Perhaps I'll post some stuff there as well.

BTW, I like your PBase photos, especially the classic SP stuff!

Sean

TCnR Wrote:
-------------------------------------------------------
> I'm sensing a difference in the three photos, more
> in the background trees, the shadows and the
> greens are different. Which is pretty minor, I do
> see great resolution in all three, nice work.
>
> Using a Dell U2312HM and firefox.
>
> Good to hear some ideas on the general subject.
> There's also great work on Railpictures dot Net,
> if you can deal with the rules and guidelines.

The PBase photos are an interesting but frustrating lesson for me, the photos with my old Nikon with prime lenses come out really sharp. Using the same scanner and digital processing but using zoom lenses, or 6x4.5 cameras, the images are soft. Pretty basic lesson learned but too late to go back.

The other lesson learned was the old saying, " f-8 and be there ".



Edited 1 time(s). Last edit at 05/27/20 16:54 by TCnR.

Having understood, and learnt quite a bit from this thread, no one has mentioned resolution. My previous DSLR took photos at 72dpi, my new one shoots at 350dpi. Is there a hard and fast rule as what resolution to upload to, or just work on the pixels as discussed. At one time, the rule of thumb was 72 for screens, and 300 for publishing. Does it make any difference for uploads? Thanks, chris uk.

1000 pixels at maximum quality.  Hopefully it posts without edit.  If you submit 2000 pixels lower quality it will now resize the lower qualtiy to 1000 pixels which you don't want.
Every compression of a jpg reduces quality as it throws out pixels.

Not on apple but seems there could be a way so you display would be withut upsizing.

Bob

Hey 55002,

DPI, or dots per inch, is really only a relevant metric for situations where you're printing photos. Even then, the number of dots an ink-jet photo printer is printing is likely far higher than the PPI (pixels per inch) specified for the print resolution in the image being sent to the printer. For example, in SuperFine mode my Epson R2880 will be putting down 5760 DPI. When I send a 13"x19" image to the printer, I might be specifying 288 pixels per inch, but the printer will then convert this to the DPI referenced above to smooth color tones and transitions, etc. This, along with the inks, helps maintain smooth tonal gradations across a wide color gamut, while preserving fine detail in the images.

In the early days of the internet, monitors had pixel densities around 72 PPI, which I believe is where the idea that 72 DPI was for the web came from. This was a time when 640x480 images would fill a good chunk of your screen, and bandwidth was quite limited. Therefore, smaller, lower quality images were needed from both a transmission and storage standpoint. Fast forward to today, and the Retina Display on the MBP I'm using has a pixel density above 200 PPI.

In the case of the resolution, 1000 pixels wide is 1000 pixels wide. Assuming a properly exposed and digitally processed image, resolution along with the "quality" settings, which is a relative measure of image compression applied in the output conversion to jpg, will determine the file size and at a high level the relative detail that can be perceived in the image. (There are a lot of "it depends" glossed over here, so my comment is a bit of a generalization.)

With respect to your previous and current DSLR, the maximum resolution image you could capture, assuming you were shooting RAW or highest quality jpg, would be equal to the resolution of the image sensor itself. So in the case of a 24MP sensor with a 2:3 aspect ratio, you would have 6000 pixels wide by 4000 pixels high. (Note this is just referring to the number of photo-diode sites and neglecting the fact that there is color interpolation due to the Bayer Pattern filter array on the sensor.) The 72 DPI or 350 DPI is irrelevant, unless it's somehow being used as a measure of jpg compression or impacting the resulting resolution of the file being saved or output from the camera. Does this make sense?

Sean

55002 Wrote:
-------------------------------------------------------
> Having understood, and learnt quite a bit from
> this thread, no one has mentioned resolution. My
> previous DSLR took photos at 72dpi, my new one
> shoots at 350dpi. Is there a hard and fast rule as
> what resolution to upload to, or just work on the
> pixels as discussed. At one time, the rule of
> thumb was 72 for screens, and 300 for publishing.
> Does it make any difference for uploads? Thanks,
> chris uk.



Edited 1 time(s). Last edit at 05/28/20 20:41 by broken_link.

There's an interesting result of my test above, however. While I generally agree with your statement, I think there's something to be gleaned from the fact that all three images, which had different resolutions and different amounts of jpg compression applied at output, all look more or less the same. Given that the images were constrained to have a maximum file size just shy of 500KB, I think this implies that there is a comparable about of information contained in each image. Therefore, whether the image is uploaded at a native 1000 pixels wide, or uploaded at a higher resolution with a lower quality output setting while maintaining the same file size and then re-sized to 1000 pixels wide by the website, there is little difference in the result. (Again, this seems somewhat intuitive given that there is ~500KB of data in all three images, even though they are 1000, 1500, and 2000 pixels wide.) While I would prefer to have control over the resize method that is used, and any additional post-resize sharpening, it just doesn't seem to matter that much in this case.

I can say that just a 20% increase in resolution to 1200 pixels makes a notable difference in the perceived level of detail on my Mac, even while holding the total file size under 750KB. A little bit more resolution can go a long way in helping your eyes fill in the blanks.

Sean

robj Wrote:
-------------------------------------------------------
> 1000 pixels at maximum quality.  Hopefully it
> posts without edit.  If you submit 2000 pixels
> lower quality it will now resize the lower qualtiy
> to 1000 pixels which you don't want.
> Every compression of a jpg reduces quality as it
> throws out pixels.
>
> Not on apple but seems there could be a way so you
> display would be withut upsizing.
>
> Bob

Hi Sean, and others, many thanks for your help, much appreciated. Cleared up several issues.  Chris UK.

Happy to help Chris.

Cheers,
Sean

55002 Wrote:
-------------------------------------------------------
> Hi Sean, and others, many thanks for your help,
> much appreciated. Cleared up several issues.
>  Chris UK.

The 72 dpi "standard" for monitors needs to be thrown away. A 72 dpi image will look bad on retina displays which are capable of much greater resolution.
RWJ

broken_link Wrote:
-------------------------------------------------------
> Hey 55002,
>
> DPI, or dots per inch, is really only a relevant
> metric for situations where you're printing
> photos. Even then, the number of dots an ink-jet
> photo printer is printing is likely far higher
> than the PPI (pixels per inch) specified for the
> print resolution in the image being sent to the
> printer. For example, in SuperFine mode my Epson
> R2880 will be putting down 5760 DPI. When I send a
> 13"x19" image to the printer, I might be
> specifying 288 pixels per inch, but the printer
> will then convert this to the DPI referenced above
> to smooth color tones and transitions, etc. This,
> along with the inks, helps maintain smooth tonal
> gradations across a wide color gamut, while
> preserving fine detail in the images.
>
> In the early days of the internet, monitors had
> pixel densities around 72 PPI, which I believe is
> where the idea that 72 DPI was for the web came
> from. This was a time when 640x480 images would
> fill a good chunk of your screen, and bandwidth
> was quite limited. Therefore, smaller, lower
> quality images were needed from both a
> transmission and storage standpoint. Fast forward
> to today, and the Retina Display on the MBP I'm
> using has a pixel density above 200 PPI.
>
> In the case of the resolution, 1000 pixels wide is
> 1000 pixels wide. Assuming a properly exposed and
> digitally processed image, resolution along with
> the "quality" settings, which is a relative
> measure of image compression applied in the output
> conversion to jpg, will determine the file size
> and at a high level the relative detail that can
> be perceived in the image. (There are a lot of "it
> depends" glossed over here, so my comment is a bit
> of a generalization.)
>
> With respect to your previous and current DSLR,
> the maximum resolution image you could capture,
> assuming you were shooting RAW or highest quality
> jpg, would be equal to the resolution of the image
> sensor itself. So in the case of a 24MP sensor
> with a 2:3 aspect ratio, you would have 6000
> pixels wide by 4000 pixels high. (Note this is
> just referring to the number of photo-diode sites
> and neglecting the fact that there is color
> interpolation due to the Bayer Pattern filter
> array on the sensor.) The 72 DPI or 350 DPI is
> irrelevant, unless it's somehow being used as a
> measure of jpg compression or impacting the
> resulting resolution of the file being saved or
> output from the camera. Does this make sense?
>
> Sean
>
> 55002 Wrote:
> --------------------------------------------------
> -----
> > Having understood, and learnt quite a bit from
> > this thread, no one has mentioned resolution.
> My
> > previous DSLR took photos at 72dpi, my new one
> > shoots at 350dpi. Is there a hard and fast rule
> as
> > what resolution to upload to, or just work on
> the
> > pixels as discussed. At one time, the rule of
> > thumb was 72 for screens, and 300 for
> publishing.
> > Does it make any difference for uploads?
> Thanks,
> > chris uk.

Image 1 is more obviously different and looks oversharpened. 2nd image is slightly better for colour than image 3. Image 2 looks best to my eyes.

Testing another image for file size vs image size and using autoresize after download.

One more...

broken_link Wrote:
-------------------------------------------------------
> Another test...
>
> broken_link Wrote:
> --------------------------------------------------
> -----
> > Testing another image for file size vs image
> size
> > and using autoresize after download.

Giving up on this soon, promise

broken_link Wrote:
-------------------------------------------------------
> One more...
>
> broken_link Wrote:
> --------------------------------------------------
> -----
> > Another test...
> >
> > broken_link Wrote:
> >
> --------------------------------------------------
>
> > -----
> > > Testing another image for file size vs image
> > size
> > > and using autoresize after download.

Additional tests

broken_link Wrote:
-------------------------------------------------------
> Giving up on this soon, promise
>
> broken_link Wrote:
> --------------------------------------------------
> -----
> > One more...
> >
> > broken_link Wrote:
> >
> --------------------------------------------------
>
> > -----
> > > Another test...
> > >
> > > broken_link Wrote:
> > >
> >
> --------------------------------------------------
>
> >
> > > -----
> > > > Testing another image for file size vs
> image
> > > size
> > > > and using autoresize after download.