Okay, all you armchair photographers...

[l_a_k_h_i_n_d_e_r]

Here is 4 stop over exp.

[l_a_k_h_i_n_d_e_r]

Some minor additions: No other adjustments other than curves is presented.

 

Also, just below the Macbeth card, in the last two scans you can see the details in a dark-green lawn-chair. Almost like the foliage in the first few scans, it is very dark. Those details are clearly missing in the first two (under exposed) scans of mine.

 

(Exposure was compensated through out for 1/3 stop, for magnification of 1:10. I could have ignored it also I guess.)

[rowland_mowrey]

James, Lakhinder;

 

To try to answer part of your questions, here is my $0.02 worth.

 

The printing process compensated for each of James prints by changing exposure times drastically in each frame. This change was the printers attempt at maintaining the overall print density at the same approximate level.

 

The apparent detail is actually there because at the same relative print density, the contrast of the image in the region of the trees is lower and therefore more detail appears to be there in the print from the underexposed negative. If the normal print was printed lighter, we would see greater detail there as well. Probably much more than in the underexposed example.

 

Lakhinder, I suggest that you repost your scans normalizing the density and balance of the 3rd neutral step from the bottom left. These results may be quite different. The reason is due to the difficulty of balancing on the darkest or lightest patch in a scene and the contrast range available.

 

Just some thoughts.

 

Ron Mowrey

[EricM]

"Some minor additions: No other adjustments other than curves is presented"

 

well, that makes the whole excerise pointless now.

[rowland_mowrey]

Eric;

 

Thanks for pointing that out. I missed it entirely.

 

Lakinder, if you decide to take my advice and adjust for the 3rd neutral step, make no adjustments but light / dark correction. No color or contrast changes. Then we will be able to do a good side by side comparison.

 

Ron Mowrey

[l_a_k_h_i_n_d_e_r]

Eric, do you know what the Fuji Frontier at Walmart did to the negatives? No curves adjusted?

 

The purpose of correcting over/under exposed film is to somehow make it look as normal as possible -- AFAIK. And as far as the film response is on the linear region, it will be easy to manipulate the curve layer in Photoshop. If I do not correct it, exactly what will be deduced from the original scan? For example, see the next two photos.

[l_a_k_h_i_n_d_e_r]

And 4 stop over exposed shot no correction:

[rowland_mowrey]

Lakhinder;

 

Try my suggestion.

 

Correct aiming on the 3rd neutral step with no contrast correction, just lighten or darken.

 

Ron Mowrey

[elliot_n]

I'm not sure what you're trying to demonstrate, Lakhinder.

 

Using an Imacon scanner, you should be able to get very good matches between the

properly exposed negative and all of the over-exposed negatives. This matching

should be done at the scanning stage.

 

If you've scanned without making corrections then of course the overexposed negs

are going to look totally washed out. And then they're going to look a little weird

when you try to normalize them by making big tonal adjustments in Photoshop.

 

I don't think your test reflects any real world situation - you need to start with decent

scans.

[l_a_k_h_i_n_d_e_r]

Elliot, there are professionals who totally bypass the corrections

in scanning stage -- and go to 16-bit capture. So that would go against your assumption.

 

Same logic applies to 8 bit scans too! I would use 16 bit scans if it were a transparency. This is obviously not. I was trying to keep the scanner software out of the loop here -- hopefully.

 

Also, what looks washed out, if it corrects well in photoshop (or however), then that is what I want!

[l_a_k_h_i_n_d_e_r]

I looked at Color Management by Fraser/Murphy/Bunting, and using the

Macbeth chart scanned values in LAB mode, used 3rd square as the reference, and made it LAB: 67, 0, 0. (See page 221).

 

So here is the series again. No other adjustment except for Brightness control, as Rowland mentioned. I could not adjust the 5 stop underexposed scan to get me the reference LAB values. So I did what I could, and left it at that. Had I adjusted its contrast, I could have reached the reference value.

[l_a_k_h_i_n_d_e_r]

Underexposed by 5 stops. Only brightness changed in Photoshop 7.0. But could not go all the way beyond the scale maximum!

[l_a_k_h_i_n_d_e_r]

Oops. the size is slightly bigger. Here again is -5 stop underexp.

[l_a_k_h_i_n_d_e_r]

-3 stop under..

[l_a_k_h_i_n_d_e_r]

2 stops over.

[l_a_k_h_i_n_d_e_r]

Here is 4 stop over, normalized scan.

[elliot_n]

Lakhinder, I still don't get it... What are you trying to demonstrate?

 

Are these raw scans? I thought that Imacon raw scans could only be processed by

Imacon's Flexcolor software. They can't be opened in Photoshop, can they?

 

If they're not raw than you haven't circumvented the scanner's software you've just

got

the settings wrong. It's like going into a darkroom, making a print of a properly

exposed negative at, say, 10 seconds, and then printing all your under and over-

exposed negs at 10 seconds. What's the point? You're going to get a bunch of very

dark and very light prints.

 

You're working with arguably the best film, scanner and image editing application in

the world - and you're producing rubbish.

 

Reala is much, much better than that.

[rowland_mowrey]

Thanks Lakhinder;

 

Now we see 3 sets. Uncorrected, 'optimally corrected' and just brightness corrected. You see the range of tonality and color that is encompassed by all of these in scanning now.

 

Printing is a similar process, optimizing the picture by 'normalizing' the exposure. The pictures James shows have some degree of 'optimization' built in making them similar to the best selectable from Lakhinders pictures.

 

But you see that the latitude of the film is about 5 over to 3 under or 2 over to 3 under. We can establish that 3 under is the max from both tests, but the overexposure varied from 2 to 5. Since overexposure is placed on the shoulder, it may be that the shoulder of the two batches of film may be quite different, or in other words the latitude of the films differ, but the ISO and EI values are the same.

 

This may be product variation or process variation. We need more tests. Heh.

 

But we already knew that negative films can be under and over exposed, and now we know that at least two stops in either direction can give us usable results. One stop is apparently quite safe.

 

This is a very good test and gives us a lot of information. I'm just sorry that I don't have any tests that go as far. I usually only tested plus and minus one or two stops.

 

Tests like this for several batches of film will reveal to us a combination of effects of process variations and product variations which will give us save exposure limits for this product. Tests like this across product or manufacturer lines will give limits and capabilities of these other films.

 

Thanks for the great thread.

 

Ron Mowrey

[rowland_mowrey]

Eliot;

 

Of course the scanner is contributing, but then the printing process contributed to James' result as well. What we are learning is how far over and under we can expose this film and get reasonable images by any means without severe losses. We are learning that there are differences in methods of testing, processing and scanning. If we keep at it long enough we will get a true benchmark for this film, not just a statement that the film is good and capable of more. We will see if it is process sensitive, if it varies from batch to batch, etc.

 

I find it worthwhile.

 

Ron Mowrey

[elliot_n]

Rowland, looking at Lakhinder's latest images most people would conclude that it is

better to underexpose Reala by three stops than to overexpose it by two.

 

Does that sound reasonable?

 

Rather than taking Lakhinder's results at face value I'm questioning his methodology.

 

Also, I think you underestimate the role of the minilab's software in James's prints. An

optical print of the underexposed negative would be much less impressive.

[l_a_k_h_i_n_d_e_r]

<i> You're working with arguably the best film, scanner and image editing application in the world - and you're producing rubbish.</i>

<br>

Elliot, I like the way you put it! And I also used one of the best camera/lens to do that!<br><br>

 

<i> Using an Imacon scanner, you should be able to get very good matches between the properly exposed negative and all of the over-exposed negatives. This matching should be done at the scanning stage.</i><br>

And if you want to focus on this aspect, then just look at my first four scans, which are actually from the same scans I show later.<br><br>

 

Rowland, thanks for your observations and suggestions. I am sure my process has flaws, as also some part of my scanner software might be out of my control! One thing I will say is, that even the original normally exposed scan (which I have not posted), does not show very bright left square block. I had not shot a normal exposure in the film I had scanned above. So I did not post. I have it from a test I did with the same film batch, but on a different day under similar light!

Maybe someday I ought to repeat this experiment a bit differently. Maybe do 16bit scans..!?

[EricM]

"Maybe do 16bit scans..!?"

 

Why would you do 16 bit scans? 16 bit is way way in the future. There's nothing that supports it for us, the best printers are 12 bit, you more than likely print on 8 bit, but weirder, you want to go through all the 16 bit effort for a 72dpi monitor?

 

If you want to do something worthwhile for this thread, take a roll of your reala in 35mm flavour, do eight bracketed exposures repeated three times over, then clip the roll at a good lab, do a front end clip at push 1, a mid roll clip at N, and an end clip at -1N. Make sure the lab clips generously and record every exposure. Photograph something tightly framed and with a worthwhile tonal range, not a white blob against a white sky with some trees in the distant background. There's some here that need to see this.

 

I'd still like to see those ten stops...

[l_a_k_h_i_n_d_e_r]

<i> 16 bit is way way in the future. </i>

<br>

Eric, do you know that Photoshop allows one to import 16bit scans.

And many photoshop books discuss it too -- if you care to read a bit about scanning techniques. <br><br>You still have not answered my question. Do you know what Fuji Frontier does to the curves?

[rowland_mowrey]

Elliot;

 

My previous post went through several revisions before I submitted it, but basically it stated what you did. James' pictures are better that one would expect, and Lakhinders are worse due to the conditions of the scan / print / test / film combinations. I don't disagree with you.

 

I still say that these are good experiments that teach us something. It does not say that one should draw premature conclusions before the experiment is over. There is more to do and this is just the entry point. Eric had a very good suggestion for the next step. Three or more rolls of film with the same exposures (over and under by 5 stops or whatever) with push and pull processing.

 

We all will learn by each step.

 

If conventional prints are made as well, we will learn how well the digital and conventional printing methods compare and what they do.

 

Lets be open minded during an ongoing process and not judge hastily.

 

Ron Mowrey

[EricM]

"You still have not answered my question. Do you know what Fuji Frontier does to the curves?"

 

I didn't read your posts. Your scanning on a frontier...

 

Of course I use Photoshop. How many filters can you use in 16 bit? Do you like using adjustment layers? What do you output on?

 

To answer your question, no. I haven't scanned nor worked on a frontier file for two years. I hope the software has been updated since.

 

If you want to work slower with larger files, then not notice a difference when printing, go fo it.

 

<a href=" http://www.ledet.com/margulis/ACT_postings/ColorCorrection/ACT-more-16bit.htm">A good read</a> on bit depth and when you should and shouldn't work in 16 bit. Make your own mind up.