Meiji 4X vs. 50/2.8 El Nikkor

[ChrisR]

Yes.
Again a thread where issues are muddling.
You're obviously right that the lens works best at about 6.7, which says something to me at least, about what diffraction does - or doesn't do.

The jpeg issue literally fuzzies things, at about the same level as the detail that we're looking at in your pic to compare the lenses.

I just compared image peeks of your fly, a fine large jpeg clip, and a ZS jpeg clip, at the same size. The jpeg artifacts are strong in your fly, absent in the FLjpeg, and I think very feint if there at all in the ZS output jpeg. So I think we're at the end of this particular blind alley!

(And my intuition, far cleverer than I am, is yelling at me to stop SPLITTING HAIRS!)

I'll delete this with apologies or others can, if posting it is breaking rules:
(Order as written)



(To get my aged PC to run at all, I loaded all the fine large jpegs from the camera into Photoshop, then, "automatically" cropped them to waste as little IQ as possible, and shrunk them, to 2250 wide, and saved as large jpegs again.
As I understand it, the Stacking progs unpack the jpeg to pixels before operating, so the degree of compression won't affect processing times, but you've already lost some IQ in the compression.)

[NikonUser]

Never any problem with posting images relevant to the discussion in any of my posts.
(Anyway, rules are expected to be broken; that's why they make the rules)

[rjlittlefield]

What we have here is the hidden assumption problem.

These images are actual-pixels from the stack, but not actual-pixels from the camera's sensor. The D2X has a 4288 x 2848 sensor, but these actual pixel images only sum up to about half that on each axis. NU writes that "The original individual frames are in both fine small jpg (causing the un-prettiness here) and RAW (NEF) files." I'll bet that "small" here means 2144 x 1424. This means that the pixels in the stacked image are effectively twice as large as we've been assuming, and therefore the effects of diffraction are only half as large.

What's funny is that when I first saw this thread, something niggled in the back of my mind about the area covered by the actual pixel crops. But I didn't follow up on it. Now I understand what niggled -- the crops cover 4 times the area they would ordinarily, working with the camera's large images.

In another thread, Graham wrote that

...or as Einstein was rumoured to have said:

"If the facts don't fit the theory, the facts are wrong"

Substitute "assumed facts", and that's a perfect description!

OK, I'm glad we have this one figured out! (Unless of course we don't! )

--Rik

Edit: to fix phrasing

[rjlittlefield]

Touching on other points raised in this thread:

1. The forum software does not modify uploaded images unless they are oversized to start with. In that case, the upload will resize and/or reduce quality. Letting it do this is a very bad idea because the quality loss is unpredictable and often painful. Keep 'em under 800x800, 200 KB, and the bits you get back are the bits you sent up.

2. We revised the rules on posting images in other people's threads a while back. It's still a no-no (without permission) in the image galleries, on the presumption that in the image galleries, the focus is on the image. But now, "In technical discussion forums, images can be posted in anyone's topic as long as they help the discussion."

--Rik

[ChrisR]

AH now I know what NU's "small" means. My word, don't you get a lot on a 16GB card Unless I turn to sports photography I'm not sure I'll ever use it

Edit - not NU's, Nikon's!

[NikonUser]

Hey! don't shoot the messenger.
I'm just following Nikon's terminology for at least their D70, D90, D2Xs.
For the D90 & D2Xs "large" = 4,288 x2,848 px; "medium" = 3,216 x 2,136; "small" = 2,144 x 1,424".

I think that if you had my computer and tried to run a 150 frame stack on ZS using "large" JPG frames and shooting to display at 800 px on a monitor you may want to consider.

[ChrisR]

Hmm, when I was a student microscopes made me see double!

when I rote:

As I understand it, the Stacking progs unpack the jpeg to pixels before operating, so the degree of compression won't affect processing times, but you've already lost some IQ in the compression.

It seems to be the Nikon "fine" that's the issue on the small files?
Presumably you wouldn't see jpeg artifacts like your fly's on a large fine?? PS does a better job.
Exposing full size takes no longer, and stacking less-compressed takes no longer.
You can run a batch program in Photoshop to run through the whole directory to reduce the file sizes.
Similarly you can check the end images to see what you might want to crop away (can be done in the same batch program), which increases stacking speed without losing any quality.
It does take some time to run the batch, perhaps 10 minutes.
That's what I do with my 4 - 5 year old homebuilt pc with 1GB ram.

I recently ran 135 frames at 2750 wide. I admit I didn't hold my breath for it though

[rjlittlefield]

As I understand it, NU's situation is that he needs to shoot hundreds of stacks. All of them need to get processed into low resolution output, smaller even than 800x800. But at some future time, he will want to process a few of the stacks -- which ones are unknown right now -- into full camera resolution outputs.

Given his camera's capabilities, the efficient way to do this is to simultaneously capture a raw NEF file and a small fine jpeg. The stack of small fine jpegs will rapidly run through any stacking program, while the stack of raw NEF files remains available for later processing at full resolution.

This strategy gives the desired output with a minimum of overhead. The only time it runs into trouble, as NU said, is when somebody ("Who, us?") mistakenly tries to use the small images for other than their original purpose.

It's worth noting, I think, that even the small images are quite adequate for their original purpose of determining which lens is better, the Meiji or the EL Nikkor. If the results had come out as a tie, then it would have been necessary to go to the higher resolution images to resolve it. But since the EL Nikkor is clearly superior even at reduced resolution, the current test answers the question.

--Rik

[lauriek]

Going back to the original comparison, it's worth pointing out if you're patient you should be able to purchase the EL Nikkor for about the same price as NU's Meiji (I had to lurk on eBay for a few weeks and I paid about $5 more for mine which is the later model, in good condition, with the nice lens tub.)

I haven't attempted a scientific comparison so far but I /think/ I prefer the output from my OM50/1.8 @ f5.6 (reversed) over the EL-Nikkor. I must get round to doing a proper comparison! (The problem is with my current setup I cannot keep the lighting in place while I change the lens, so this is tricky!)

[rjlittlefield]

HERE, ChrisR commented that some of the actual-pixel crops posted earlier in this thread contain a lot of what look like JPEG artifacts.
His example is this:



I agree, these do look like JPEG compression artifacts.

Going back to the actual-pixels crop from which these were taken (first of four HERE, I believe), what I see is a 799 x 800 pixel image packed into only 51.9 KB. That's a pretty high level of compression.

I would be surprised if the D2X did this in "fine" mode. ZS might do it at quality setting 9 or below (max=12). At least, that's what it takes to give close to the same size, using that 799 x 800 image as input. Or it could have been done by Photoshop using a quality setting that is suitable for packing much more detailed images into the forum's 200 KB limit.

--Rik

[ChrisR]

The image two posts above that "top" one, is 800 x 531 (here) , compressed only to 199.9 kB, and that one doesn't show much in the way of artefacts, so indeed the effects seem to be post-camera.

This strategy gives the desired output with a minimum of overhead. The only time it runs into trouble, as NU said, is when somebody ("Who, us?") mistakenly tries to use the small images for other than their original purpose.

I am not wishing to criticise NU's workflow at all, I hope I didn't convey that impression. I was simply curious about a couple of aspects which cropped up in the analysis of the pictures. I was under the impression that these had been taken specifically to look at the sharpness comparison between the two lenses, and apertures.
In a minor revelation, close inspection of the ORIGINAL 800pixel crops, in the first post of the thread (800 x 400, 200kBytes) does NOT show the jpeg effects, so NU's method hasn't "run into trouble"! I was misled by their appearance later on, not being aware that they'd only just been introduced!

I do feel sorry for NU in all this, There he goes usefully posting a comparison between two highly relevant lenses, and he comes under a barrage on matters outside his point, which was well settled in the first post!
The Barrage is certainly not intended as criticism, and I hope I'm not the only one who has gained some useful insight from the exercise.