You are here

13 posts / 0 new
Last post
exporting out of aperture to plugin in 16bit or 8bit? #1
stump4545's picture
by stump4545
September 27, 2012 - 10:40pm

is there much lost in exporting to silver effects pro in 8 bit rather then 16bit?

for black and white images or color images same thing?

just that 16bit really slows down my imac compared to speedy 8 bit.

thank you.

stump4545's picture
by stump4545
September 28, 2012 - 11:23pm

sorry, just so i understand perfectly, if i want to continue to use silver effects pro, exporting to SEP in 16bit will give me the best quality?

if i export to silver effects in 8 bit what am i losing?

thank you for all your help.

PhotoJoseph's picture
by PhotoJoseph
September 29, 2012 - 8:48am

Stump,

haha you’ve started a discussion here and not quite gotten the answer. I don’t think there is an easy answer though, except for, “it depends”.

On a purely technical level, your camera shoots RAW at 12 or 14 bit most likely. You bring that into Aperture which works in a floating point 16-bit space, meaning as Grant explained, that all edits aren’t actually rendered to the file until you output it, so as long as you’re in Aperture, you’re working in the best possible quality.

An 8-bit TIF generated out of Aperture will CLIP any data that’s out of range. That’s extremely important to understand. So if you adjust the levels and curves such that ALL DATA is within the 8-bit range, and you export as 8-bit, there will be no loss to the file. Think of it like cutting the top off a tall glass that’s only half-full of water. You don’t lose any water, you just decreased the size of the container. On the other hand if you cut the top off a glass that’s full of water, then you will leave water on the table. Same thing with the file.

To continue the analogy, if you take that 8-bit file [shortened glass] and try to fill it with more than 8 bits of data [more water], you can’t. However unless you’re doing something really extreme to the image, that’s not likely to be a problem.

For B&W work, my thoughts are that it really won’t matter. But then again I did do 16 bit “just to be safe” for my big photos. Maybe you should just try it yourself. Using the same photo, open one in 8 and one in 16 into SEP2, apply the same extreme adjustment, then compare the final output side by side, and to really compare, stack ‘em in Photoshop and run Difference.

@PhotoJoseph
— Have you signed up for the mailing list?

gfsymon's picture
by gfsymon
September 29, 2012 - 1:20pm

Joseph,

that’s not quite accurate.

Aperture should not be clipping values during export. This is why we have Colorsync. It will ‘re-map’ them into the chosen colour space using what we hope, is the best ‘intent’ for the purpose. We can’t choose the intent for export in Aperture, unlike PS or even the MacOS.

So … if for example an image is in AdobeRGB and contains colours which lie outside of sRGB, when you choose to export using sRGB, those colours, lying outside sRGB will be ‘mapped’ to new colours which lie inside sRGB. The result of this will be a loss of saturation, because it is the saturated colours which AdobeRGB can contain, that are outside sRGB’s gamut. However … there should be no ‘clipping’. Nothing is getting ‘chopped off’. Instead the file is being altered in the best way possible, to retain the visual perception of the original AdobeRGB file, but in the smaller sRGB space (perceptual intent).

This though does not preclude banding issues appearing, although most commonly banding that shows up in an 8-bit file may well have been present in the 16-bit version, but imperceptibly. The transition to 8-bit will exaggerate the effect and it will become a problem. That said … banding is not a major issue these days. It used to be though.

So the ideal thing to do if you want to work in 8-bit, is to do any big tonal adjustments inside 16-bit and then export to 8-bit and do the rest of the work on the file.

(I will re-iterate though, that you can do anything that Silver Effects can do, in Aperture itself.)

PhotoJoseph's picture
by PhotoJoseph
September 29, 2012 - 2:26pm

Grant,

Hmm… I’m thinking beyond color gamut though. If I have blown highlights for example, but I shot RAW, chances are I can recover some of that detail, even though it’s out of the 8-bit range, because I can pull data back into range and see it. However if I convert that RAW file to an 8-bit TIF before pulling those highlights back in, aren’t they clipped? Are you saying that this info would be compressed into the 8-bit space on export?

@PhotoJoseph
— Have you signed up for the mailing list?

PhotoJoseph's picture
by PhotoJoseph
September 28, 2012 - 4:00pm

To my knowledge, the only advantage is if you’re either not bringing the full data within range of the histogram to begin with, or if you plan to push out to out-of-gamut colors in the process.

If you’re careful to bring all data into range of the histogram to begin with, it shouldn’t even be an issue.

If you don’t, then you’ll clip when you convert, and that’s the same as shooting JPEG in camera and not exposing correctly. Once it’s gone, it’s gone.

@PhotoJoseph
— Have you signed up for the mailing list?

gfsymon's picture
by gfsymon
September 28, 2012 - 4:00pm

Even though you’re in black and white, it’s still a question of the number of ‘steps’ that you have in your file. If you are doing any kind of big curve adjustments (tonal) then 16 will make a big difference and avoid any banding.

stump4545's picture
by stump4545
September 28, 2012 - 4:37pm

so i guess to ensure the best quality and tonal range it is good to export in 16bit to silver effects and just deal with the added processing time.

gfsymon's picture
by gfsymon
September 28, 2012 - 4:44pm

To ensure … yes.

It’s a difficult call, because 8-bits covers everything that human beings can distinguish and often 8-bits will be plenty to cover your adjustments and keep transitions smooth … but what if it isn’t ? You can either spend time scrutinising your finished files, to ensure that all is good … or just use 16-bit and not have to worry. Actually … you can push 16-bits too far as well, with extreme adjustments, but these are well outside normal usage … but it does bring back the same question … is 16 enough ?  :) :)

PhotoJoseph's picture
by PhotoJoseph
September 28, 2012 - 5:09pm

Hm, not totally sure that I agree, but I could be very wrong here. Our screens are 8-bit, so it’s impossible to see 16-bits of data at once on screen (without one of those super fancy/expensive HDR screens, at least). Then there’s print; I know that 16 bit printing is available and maybe it makes a visible difference in color, but can it in B&W? And, can anyone really tell the difference without a scope?

I’ll be honest, I did my recent B&W gallery hanging (see The B&W Photo and go to Gallery) as 16-bit files because I didn’t have time to test any theory and frankly, I figured “what the hell”. But it may have been a complete waste of time going that big.

What we need in here is some concrete experience of 8-bit vs 16-bit, in color and B&W, all the way through to print.

Anyone? Anyone? Bueller? Bueller?

@PhotoJoseph
— Have you signed up for the mailing list?

gfsymon's picture
by gfsymon
September 28, 2012 - 10:20pm

Hell, what do I know Joseph.  ;) ;) I’ve been working in 16-bits all the way back to 1997 in LP. 8-bits out, but all 16-bits right up until the last push of the button for build (export).

PS used to get very ugly back in those days in 8-bit. Simply the act of opening a file in PS reduced it to 7-bits. A single sharpening would bring it down to 6-bits and a curve adjustment on top would reduce it to 5-bits. Yuk. These days PS only applies edits at the end, when you flatten the file, which really copied the LP approach and significantly limits the damage. Aperture is much like LP in that all internal operations are in 16-bit and your edits are all applied at the end, when you export, so it gives you the cleanest file possible.

I guess to answer ‘stump’ … knowing that silver effects pro give you nothing that you can’t do in Aperture alone, the answer would be to learn how to use Aperture more throughly, and enjoy not having to export a 16-bit tiff ? It’s all just tonal adjustments. Curves will do pretty much everything … but I have to admit, that the new Highlights and Shadows brick makes a lot, very easy.

gfsymon's picture
by gfsymon
September 30, 2012 - 8:28am

Joseph,

clipping is very bad and no body wants it, so software engineers go out of there way to avoid it. It’s like your glass analogy or perhaps like using scissors to cut around an image. Bad news.

Re-mapping, will either:

a) Squeeze colours into the smaller gamut. Think ; digital ‘cyan’ skies a few years ago. These were due to clipping. Usually when photographing a blue sky, the red the channel hits 255 first, leaving the other 2 alone to provide colour information. As those 2 remaining channels increase in exposure towards 255, they increase their influence in the relationship between the 3 channels, over the rendered colour. Eventually, one of the other 2 remaining channels will top-out at 255 and the colour will once again, veer towards the remaining colour (usually green) and so you will again see a weird transition, until finally the 3rd channel will also top-out at 255 and you will just have white (255/255/255).

The red, then blue channels were ‘clipped’ so the colour was inaccurate. Eventually manufacturers and software engineers learned to make this transition to white 255/255/255 more gracefully and so nowadays software re-maps at the white end, to simulate the way that film would blow highlights.

b) Re-map more literally into a smaller space (or larger space!!) so that the perception of the image/colours is as close as possible, but using subtly different colours. This can be done either 1) perceptually, so that the colours ‘look’ as similar as possible to the original, or 2) relatively, so that the relationship (their physical position inside the colour space) remains the same.

For your raw example :

The software (Aperture) is giving you complete control and so is allowing you the choice concerning the blown highlights to either a) leave them ‘clipped’ (see above) or b) manage the extent of the remapping yourself. But NB, this is not ‘recovering’ anything, because there is nothing there to be recovered. Instead, it is using detail information available in the remaining 1 or 2 channels to ‘create’ information for the clipped ‘missing’ channel. I’m not sure precisely how this is done, perhaps it varies between softwares, but I would guess that it looks at surrounding pixels for blown highlights (those at 255), then looks for that channel’s surrounding pixels which are still at 254 or 253 and then compares the relationship of those 254/253 pixels to the other 1 or 2 channels. It then knows what colour it should be. Now it can squeeze all 3 channels together (remapping) towards a 254 max, but using the colour relationship that it now understands and thus giving the correct colour (for the scene) and … it will avoid banding(!) and … will have as natural a look as possible.

That’s no small feat and as you will have noticed, the more colour information that you have lost, the harder it is for the software to ‘create’ it.

Dennis's picture
by Dennis
June 23, 2014 - 1:40am

Hi,

I like to add the following question to this discussion.

When making sure there is no clipping before exporting the image as an 8-bit TIFF, does this limit the external software in color usage because the TIFF container is 8-bit? In other words, is it possible for the external software to generate (color) banding because of the 8-bit TIFF container?

With kind regards, Dennis

IDEMI Photography | www.idemi.nl

You may login with either your assigned username or your e-mail address.
Passwords are case-sensitive - Forgot your password?