How To Change 8 Bit To 16 Bit Photoshop

#1

Posted 23 April 2012 - 01:21 PM

For years I've used Eddie Trimarchi's great fitsplug routine to transfer my 32 bit IEEE format fit files from MaxIm to Photoshop's 16 bit format (CS3). Now I've moved on to the 64 bit version of Photoshop (CS5). I upgraded to the latest version of fitsplug to get 64 bit capability. The new version brings the data in with Photoshop's 32 bit format. This, by itself is ok. I used to have to apply a series of curves/levels to convert a largely black astro image to something useful. Apparently Photoshop does something similar automatically with 32 bit images since now I get largely processed image to start with. However, there isn't much I can do beyond this without converting the data into a 16 bit format. When I try to convert to a 16 bit format I get the HDR Toning screen. With some images I can find a combination of settings that work ok and then I can continue processing in the 16 bit format. However, this HDR Toning screen totally destroys other images and a I can't find any setting that prevents this destruction.

I get a similar destruction if I try to apply 2 iterations of HDR Toning to an image, so perhaps I'm getting a "free" HDR Toning when I first bring in the 32 bit image and another when I convert to a 16 bit format. Interestingly I get a totally different behavior if I convert a 16 bit image to 32 bits rather than open a 32 bit file with fitsplug.

From my discussions with Eddie, it seems that most people using 64 bit Photoshop with his plugin convert the data to 16 bit format and then continue normal processing. But, I'm at a loss as to how to do it. How do I convert to 16 bit format without destroying my images?

I posted this question on ccd-newstro a few days ago but no one could help. Hoping someone here can.

#2 vpcirc

Posted 23 April 2012 - 03:02 PM

You need do two stretches in Maxim first. First do a log-max pixel-unlimited. The second stretch to 16 bit is: screen stretch-linear only-16 bit.I think I have that correct. I can't find my link to IP4AP S2N link to double check. I think also if you perform a ddp, it converts it to 16 bit, but as I said I can't verify that without Warren's tutorial to refresh my memory. Here's the link to the site, there's a free one on Maxim, it may cover that since its on calibration/combine/stack
http://www.ip4ap.com/soup2nuts.htm

#3 astronutcase

Posted 23 April 2012 - 03:22 PM

Yes, I know I can stretch the image in MaxIm and save the result in a 16 bit fit or tiff format. That bypasses the 32 bit Phortoshop problem and is the way many imagers like to operate. However, I'm trying to take advantage of the 32 bit fit to Photoshop capabilty of fitsplug. I understand many people do operate this way - I just can't figure out how.

#4 vpcirc

Posted 23 April 2012 - 03:58 PM

Sorry, I misread the question, here's a link
http://www.imagingde..._Streaming.html

But you'll notice he has pre-processed and is in 16 bit prior to HDR work. Ken is a pro.

#5 astronutcase

Posted 23 April 2012 - 04:09 PM

Thanks, I viewed Ken's video as one of my first steps in learning about HDR Toning. However, he doesn't talk about the issue I'm seeing. Right now, I'm using 16 bit images since I can't figured how to get around the problem. This is actually a backwards step since the old fitsplug running with 32 bit Photoshop allowed me to input a 32 bit IEEE formated fit image.

#6 vpcirc

Posted 23 April 2012 - 04:37 PM

Don Goldman says doing your stretching in photoshop is the old way that most imagers no longer do. There's too many great programs that pre-process the image to eliminate the need. Goldman's presentation is from AIC 2010, so I'm not sure if that's available. I personally don't know of any advantage to processing in fits mode anyway. I have the fits plug, but I rarely open an image in fits mode anyway, but I'm using CCDStack which has a wonderful tools that get the image very close prior to photoshop. Maybe there's some advantage in photoshop that I'm unaware of. Ken is in 16 bit mode and using HDR to enhance certain details. I can open a 32 bit fits in 64 bit PS using the 64 bit plugin.

#7 astronutcase

Posted 23 April 2012 - 04:47 PM

When you open a 32 bit fits in 64 bit Photoshop do you get a modified image (as I do)? I wonder if part of the issue is due to a 32 bit IEEE fit format rather than 32 bit binary fit format. I've been using the IEEE version. I'll have to see if the same thing happens with binary.

#8 vpcirc

Posted 23 April 2012 - 06:11 PM

I always save my masters in IEEE, but even at IEEE, I don't think my image would be greater than 32 bit unless I Did a log-max pixel- unlimited stretch only. I asked a pro and the response was "does PS alter/stretch his supposed 32-bit FITS or he doing something in Max but never saw the change when he worked in 32-bit"

#9 astronutcase

Posted 23 April 2012 - 06:31 PM

It's definitely Photoshop that's "stretching" the data (or something like a series of curves and levels). Since it doesn't do this if I load a 16 bit file then convert to 32 bits inside Photoshop, I suspect there may be some kind of option or setting that should or shouldn't be set that I don't know about.

#10 vpcirc

Posted 23 April 2012 - 07:31 PM

when I open a 32 bit fits image in photoshop, when I look under mode, it says 32 bit, so I can't see how photoshop is doing anything to the file.

#11 astronutcase

Posted 24 April 2012 - 06:28 AM

This is strange. When I open a 32 bit fits image I see an image that has an obvious non-linear stretch over the image I see in MaxIm (or what I see in Photoshop if I read in a 16 bit version). It looks like its gone through HDR Toning.

#12 vpcirc

Posted 24 April 2012 - 08:31 AM

It has to be in Maxim, not photoshop. Do you have auto stretch checked when you save the file?

#13 astronutcase

Posted 24 April 2012 - 10:17 AM

No, I never use auto stretch. I've tried it without any stretch, with DDP, and with manual stretch in MaxIm. They appear even more stretched in Photoshop with 64 bit formats and essentially the same as in MaxIm with 32 bit formats (or as close as one could expect). If this doesn't happen to you then there must be some Photoshop setting that essential says "autorun HDR Toning on load" or something is broken. CS6 should soon be out and I have a free upgrade so maybe that will resolve my problem (cause I don't know where else to look)

#14 astronutcase

Posted 28 April 2012 - 06:04 PM

Ok, I have the answer. I just got back from the imaging conference (NEAIC) where I asked every expert I could corner about this issue. Alan Erickson is the astro guy at Adobe. He provided a great explanation of what appears to be happening. Photoshop assumes 16 bit data has a gamma of 2.2 the same as a Windows display. However, Photoshop assumes 32 bit data is linear or has a gamma of 1. To display 32 bit data it has to raise the data's gamma on the display from 1 to 2.2 - a significant increase. Depending on how bright the astro image is the user might not notice the change or might see a significant improvement or might (especially if DDP was used) see the image destroyed. I've seen all 3 of these cases. This display brightening isn't necessary when you convert a 16 bit image to 32 bits since the display gamma doesn't need to change - again exactly what I've seen. HDR Toning can be a useful option but Alan wasn't sure why it's necessary when converting from 32 bit to 16 bit. The normal local adaption method in HDR Toning can again radically change the data. However if you change the method to exposure and gamma this destruction goes away (at least in the images I used).

PS: the imaging conference was GREAT (and not just because I won a $2450 FLI Ceneterline filter wheel door prize)

#15 Rubikean

Posted 08 June 2014 - 11:47 AM

I know this post is a couple years old, but thank you for sharing this explanation. I was having this same situation and having trouble locating the problem in our process. This was exactly what I needed to know. Thanks again.
-Chris

#16 blueman

Posted 08 June 2014 - 01:09 PM

The one thought I have on this is the camera puts out 16 bit data. Then it is converted to 32 bit floating point by some programs and processed. But in the end you either go back to 16 bit Tiff or 8 bit JPG.

So, what is the benefit of all this changing back and forth? I work in 32 bit floating point in CCDStack and then go back to 16 bit tiff to work in Photoshop CS2.

Blueman

#17 alpal

Posted 08 June 2014 - 06:07 PM

The best way is to stack the images in say DSS & save the result as a 32 bit FITS file.
Import this into NASA FITS Liberator. ( a fantastic free program )
Put the sliders from the lowest to above the highest value -
then stretch with high compression such as x^1/5.
Then save as a 16 bit TIFF for Photoshop.

The reason is that even though the camera supplies 16 bit files -
you can get an extra bit or 2 out of stacking.
That is where the high noise information is so that will help to stretch out & show
the otherwise invisible halos of galaxies etc.
It will also help to compress the entire dynamic range down to
something that is workable in Photoshop.
Remember of course that your final image will be RGB on a computer screen & only 8 bits.
It's all about compression of the data to end up with 8 bits.

cheers
Allan

#18 View2

Posted 15 August 2016 - 04:17 PM

The one thought I have on this is the camera puts out 16 bit data. Then it is converted to 32 bit floating point by some programs and processed. But in the end you either go back to 16 bit Tiff or 8 bit JPG.

So, what is the benefit of all this changing back and forth? I work in 32 bit floating point in CCDStack and then go back to 16 bit tiff to work in Photoshop CS2.

Blueman

Thank you so much for this post it's what I needed to know. When using deep Sky stacker my output is in 32bit rational format. Photoshop cannot do Curves in 32bit so I have to change to 16 bit to manipulate Curves in Photoshop. Many of the settings in the HDR toning conversion window will desrtroy the image. I have only been able to use the gamma and exposure setting and sometimes photo realistic. This was a bottleneck in processing I am happy that I can simply work with gamma and exposure and adjust my color curves afterwards.

Edited by View2, 15 August 2016 - 04:18 PM.