Converting 8 bit to 16 bit

[DuFrank]

Hi there! Is there any use in converting an 8 bit image to 16 bit or is it just as useless as converting a low quality jpeg to a highres tiff, and expecting the same quality as a original highres tiff. Hope anyone can make it clear to me?

[Paul C]

None whatsoever - you gain no image data.

pbc

[DuFrank]

Thanks Paul!

[jimhoerricks]

Modern equipment is capable of capturing between 10-14 real bits per channel. In many cases there is a lot more detail in the image than you can actually see on the screen - shadow and highlight detail for example. Having that large spread to work with allows you much more freedom to discover and retain that detail with little risk of clipping out that detail.

I script the opening of my images to convert to 16bpc and ProPhotoRGB. This gives me the largest space to work with and assures that (since I'm going from a small space to a large space) nothing is clipped. In my workflow, it is essential that all available image data is retained. When I am ready to deliver a print, I can step the image down carefully - focussing on the preservation of important details. In this way, I can produce the best possible print as I have as much control as possible from start to finish.

I hope that this helps. If you are interested, Dr. John Russ has some good books on imaging that really go into this in detail.

[Paul C]

In the case of a 12-bit Camera raw file I will choose to open it as 16-bit instead of 8 in order to retain as much image data as possible. But converting from 8 to 16 gains you nothing.

pbc

[kialua]

For my client photo shoots using raw, I open them to 16. After I finishing any work on them and if I need to send for actual photos, the professional labs here only take them at 8 bit jpgs so I convert them and save a copy for that. I always found that odd.

jimhoerricks welcome. Your field looks very interesting and waaay over my head.

[telefan]

I had my negatives scanned at a local lab. They're 8-bit, 4535 x 3035 and were saved in BMP format. First off, are there any concerns here, assuming I will want to use these in some professional capacity some day (but not sure how)?

Also, I have a book that says to do the following at the beginning of the workflow and I'm wondering if someone could commment on whether this is good, bad or unnecessary.

Open 8-bit image
Convert to 16-bit (image --> mode)
Image -- Image Size
Change unit of measure to percent and type 50 in width or height
Ensure resample image is selected and click OK.

The book implies that the resulting image is now a 'true 16-bit' image. That may be an exaggeration but I'm wondering if I should work with this '16-bit' image or the original 8-bit scans as my starting point? Lastly, would it just be best to have them rescanned as native 16-bit?

Thanks!

[kialua]

What program is the book's instructions dealing with?
What size are the photos to begin with that it tells you type in 50" in length and width?

My hunch is if it is a digital photograhy software, it is not dealing with professional printing press industry standards. I use 16 bit but it doens't always let me do work on it in Photoshop, in which case I change it to 8 bit but I have never changed it up to 16 afterwards.

Anyone else?

[telefan]

The instructions are for Photoshop. It says to type 50 in the length OR width as a percent, not absolute value. They're specifically saying to do this before editing the brightness/contract to reduce data loss. We know that brightness/contrast isn't the right tool to use but from what I've read everywhere, whatever edits Photoshop can do in 16-bit, should be done before stepping down to 8-bit. If I follow those instructions, it makes the pixel dimesions smaller (thus producing a 16-bit image???)

[Paul C]

Every time you convert modes, bit depths, change image size, etc you are altering (and probably degrading) image data. What is to be gained by the above method? Nothing in my opinion.

Your master asset is as good as it's gonna get. Any tampering will alter that and if you decide to edit the master it should be resaved as a different filename reflecting it's current state.

You are not going to gain anything by reducing the pixel count and changing to 16-bit so I really don't see the point. It's not like there are editing procedures that are only available on 16-bit images. You are also usually going to go back to 8-bit eventually for an output device.

pbc

[telefan]

Thanks Paul, that makes sense to me. There's obviously a lot of misinformation out there, even in printed books. I'm just getting into this and I'm trying to come up with a workflow by cross-referencing infomation I'm gathering.

[The Repro Kid]

Wow, this is getting interesting. Paul there are lots of 3D and motion applications that can benefit from 16 bit images whether the images begin as 8 bit or not. When converted to 16 bit they can be used to generate ultra complex rendering backgrounds. It doesn't mater what the color is so much as the applications need the extra information to generate the HDRI (http://en.wikipedia.org/wiki/High_dynamic_range_rendering) complexities.

I had my negatives scanned at a local lab. They're 8-bit, 4535 x 3035 and were saved in BMP format. First off, are there any concerns here, assuming I will want to use these in some professional capacity some day (but not sure how)?Okay, what professional use? If it's printing then 16 bit is probably overkill, as paul is pointing out. HDRI rendering and other similar things are another story. What is your End Use?


Also, I have a book that says to do the following at the beginning of the workflow and I'm wondering if someone could commment on whether this is good, bad or unnecessary.

Open 8-bit image
Convert to 16-bit (image --> mode)
Image -- Image Size
Change unit of measure to percent and type 50 in width or height
Ensure resample image is selected and click OK.

The book implies that the resulting image is now a 'true 16-bit image.'

Yes and no.

Yes it's true 16 bit, that happened on step two, but only by definition, not through the assumed benefit of changes in scale.

Doing all the Hocus Pocus with the Image size changing will yield the same result, whether you convert to 16bit or not. What happens is, changing the image size to 50% compresses the image. This "Boosts the levels." Depending on how poor your original image is, this may help to manipulate things like contrast and brightens (never use!) becuse you have compressed the levels together, so you now have some room to spread them apart, which is what contrast and brightness does. But if you open up your histogram palette and study the changes as you go through the resizing moves, you will find that the histogram changes are identical, whether you change the image to 16 bit or not (this is paul's point). So the only reason to do all the image size Hocus Pocus is to compress your levels so it is not as easy to totally destroy your image levels with the contrast and brightness controls are used.

That may be an exaggeration but I'm wondering if I should work with this '16-bit' image or the original 8-bit scans as my starting point? What is your End Use. If your end use truly requires 16 bit then you should work the 16 bit conversions. HDRI Background imaging comes to mind, is that what you are making?

Lastly, would it just be best to have them rescanned as native 16-bit?That is only case where I'd call it a "True 16 bit" because your histogram will be far different from the histogram of the converted image. "Native Format" is not the correct way to understand the difference between an 8 bit and a 16 bit image. The difference is in the ultimate amount of information yielded through a difference in the gathering method. The way the information is gathered and stored results in a greater yield of usable information.

[telefan]

I'm approaching this as a photographer so print would be the final output.

[Paul C]

Didn't know that about 3D programs Repro.

I am an advocate of a light touch - do as little as you can to the masters to achieve the results you want - if the photos are good in the first place you don't need to do very much at all. NEVER alter the masters - always work on a copy - you never know how you might need to repurpose these in the future.

The main thing is that you get enough res in the first place so you don't have to upres. This will go a long way to preserve the integrity of what was shot.

I had a large format slide scanned that yielded a 3/4 Gig file (can't tell you the pixel dimensions). This had to be repurposed in so many ways that I got it scanned as high as I could - if I remember correctly it was at 9000ppi. Can you say film grain?:D It was a very good decision as I used that for a year in all sorts of configurations.

pbc

[kialua]

I use professional photosgraphers on every project I do and the end purpose is printing also, 4 color process offset printing. Do you mean printing press printing or printing to produce a photograph? Just wondereing.

I'm approaching this as a photographer so print would be the final output.

[The Repro Kid]

Didn't know that about 3D programs Repro.

I am an advocate of a light touch - do as little as you can to the masters to achieve the results you want - if the photos are good in the first place you don't need to do very much at all. NEVER alter the masters - always work on a copy - you never know how you might need to repurpose these in the future...

pbc
Well sure but here's the deal. Changing an image from an 8 to a 16 bit res photo doesn't "up res" anything, it just changes the availability of information. For HDRI, information is now available that was not before even though nothing has changed, It a restructuring of data availability, not data. If the HDRI application needs data to perform it's task, now it can find it (not with printing apps however), regardless of how crappy the photo is.

What is comical about the "book tutorial" is the cheese ball maneuver of compressing levels to lessen the impact of the Brightness and Contrast adjustment.

[The Repro Kid]

...Do you mean printing press printing or printing to produce a photograph? Just wondereing.Maybe if you could clarify with some details, telefan. I'm too am still unclear.

that being said, if you are a photographer and you want your images to be in 16bit, then scan them in as 16 bit. do not convert them from 8 bit thinking you will get 16 bit, regardless of said imaging hocus pocus you may have read in book.

Not that I don't like books :)

Can you say Dark Ages?