1

I was under the impression 16 bits would keep the same depth. But for my great shock I have noticed that there is a considerable lost of depth.

Testing with a Canon Mark II raw and using both Lightroom and Camera Raw exporting to TIFF or PSD 16bit. And yes, I know how to set camera raw to produce 16bits instead 8

Notice how with RAW the Sun is kept when reducing exposure with Photoshop but is lost with TIFF. If fades directly to a bland dark grey without any sun shape detail.

Include the link to the original RAW and the TIF in case anyone wants to check. And if you want to reproduce the problem simply load it in photoshop (or lightroom) and save it as 16 bits TIF. https://www.dropbox.com/sh/arsg3eem4dllc0m/AACM-Mdqy_8BNw8GjPJ6dEvma?dl=0 enter image description here

I add more examples to explain my point that there is a evident degradation of data. About that the sun is overexposed, I talk about dynamic range lost compared with the original RAW as shown in the example. This is part of a multi bracketed series of photos and I have other exposures. You cannot capture the full range of the sun with a single exposure. Example with RAW: enter image description here Example with RAW moved to Photoshop and abandon camera raw, in 16 bits, 0-1-2-3:

enter image description here

Now I convert it to 8 bits to see the difference. Again 0,-1,-2,-3 exposure. And the results are identical to the 16 bits no RAW. And this certainly should not happen. Again you have the original RAW file in my link if you want to double check. enter image description here

It would be less of a problem if I would not get same results in Lightroom and camera raw, leaving me with no alternatives. I hope everybody will agree 8 bits and 16bits should not look so similar, unless there is something wrong here.

  • you can export to PSD and then save PSD as OpenEXR if needed. – skyde Aug 22 '18 at 21:24
3

The 16-bits used to record raw data and the 16-bits (per color channel) used to record a demosaiced and gamma corrected TIFF or PSD are not used to represent the same exact thing in the same exact way. Expecting a 16-bit TIFF to be the same as a 16-bit raw file is a bit like expecting a 16-bit WAV audio file to be the same as a 16-bit WMA file. They both contain information about the same sounds, but how much information they contain and the way they represent that info is very different.

In the case of the raw data, the 12/14/16 bit information is of a single linear monochrome luminance value for each pixel well. Effectively what we have with a raw file from a Bayer masked sensor is a B&W photo (but with a linear response - more on that below) using a green filter for 1/2 the sensor's pixels, and blue and red filters respectively for 1/4 of the pixels.

To get color information from a raw file, it must be demosaiced. Again, back to our B&W film analogy, by comparing the differences in brightness of the same areas in three B&W images of the same framing with each filtered respectively for red, green, and blue the three monochrome images can be used to create a color image. That's the way most color astro photographs through the mid to late 20th century were produced. That's the way color film works. It's the way digital sensors work. It's the way our retinas + brains work.

Before we convert raw data to a TIFF, we also usually apply gamma correction (a light curve that converts the linear response in the raw data to a logarithmic response closer to our human perception - again, we use the same words but it is not exactly the same thing as when we apply gamma correction to the signal going to a CRT display). If we demosaic to produce color information and then translate the numerical values from a raw file to the TIFF image format without applying gamma correction, we get something that looks like this:

enter image description here

With gamma correction applied, the same image looks like this (the low-res embedded jpeg preview from the raw file used to generate the TIFF above):

enter image description here

Keep in mind that when you 'open" a raw file on your computer or other device, you're not looking at the pure, unadulterated raw data on your screen. You're not even looking at a downsampled version of all of the information contained in the raw data. Not even close! What you are seeing is either a jpeg preview image generated by the camera and attached to the raw data file when the photo was recorded or you are seeing a similar rendition of the raw data generated by the viewing application. In either case, the raw data has been highly processed to make the information viewable in a way that we expect. That processing must make certain decisions about black point, white point, gamma correction, white balance, etc. When we "view" a raw file we are looking at one of countless possible interpretations of the raw data. That is why we can edit raw files non-destructively. When we move all of the sliders around and click various buttons we are just telling the viewing/conversion application how we want it to reinterpret the data in the raw file.

Once the data in the raw file has been transformed into a demosaiced, gamma corrected TIFF file, the process is irreversible.

TIFF files have all of those processing steps "baked in" to the information they contain. Even though an uncompressed 16-bit TIFF file is much larger than a typical raw file from which it is derived because of the way each stores the data, it does not contain all of the information needed to reverse the transformation and reproduce the same exact data contained in the raw file. There are a near infinite number of differing values in the pixel level data of a raw file that could have been used to produce a particular TIFF. Likewise, there are a near infinite number of TIFF files that can be produced from the data in a raw image file, depending on the decisions made about how the raw data is processed to produce the TIFF.

The advantage of 16-bit TIFFs versus 8-bit TIFFs is the number of steps between the darkest and brightest values for each color channel in the image. These finer steps allow for more additional manipulation before ultimately converting to an 8-bit format without creating artifacts such as banding in areas of tonal gradation.

But just because a 16-bit TIFF has more steps between "0" and "65,535" than a 12-bit (0-4095) or 14-bit (0-16383) raw file has, it does not mean the TIFF file shows the same or greater range in brightness. When the data in a 14-bit raw file was transformed to a TIFF file, the black point could have been selected at a value such as 2048. Any pixel in the raw file with a value lower than 2048 would be assigned a value of 0 in the TIFF. Likewise, if the white point were set at, say, 8,191 then any value in the raw file higher than 8191 would be set at 65,535 and the brightest stop of light in the raw file would be irrevocable lost.

It appears that when you converted your raw file of the sun into a TIFF, your white point was set considerably lower than the maximum value of the raw file. Everything brighter in the raw file than the selected white point has the same value in the TIFF, so no detail is preserved. Reducing the brightness of the TIFF file just results in the highest values being shown as darker shades of gray instead of white, but they will all still be the same shade of gray. The differences between pixels with values higher than the white point used to create the TIFF were discarded when you created the TIFF.

For more about how and why color information is lost when more than one color channel is at full value, please see: Why is it that when the green channel clips, it turns into blue?

  • I didn't edit the raw file, simply loaded it and exported it as Tiff 16bits. Michael Clark, we are going to suppose you are right. How we solve the conundrum of saving an edit from lightroom without losing data. For example I edit it in lightroom to remove noise and chromatic aberration before sending them to a HDRI program (as EasyHDR). But it becomes evident that even I gain those edits I lose color depth. In comparative if I send directly the RAW to EasyHDR I get a better color depth (I can export to EXR keeping a lot of the sun) but I lose the edits. The problem would be solved if lightroo – Vic Sha Mar 8 '18 at 13:55
  • Agree. Michael, it seems your response is a more detailed version of "ths", basically Tiff 16 will cripple the exposure no matter what, from RAW. The only possibility to keep it is to toned it, losing dynamic range, that is what I want to keep for 32 bit creation. The problem is that I need it to create 32 bits bracketed panos and I dont want to tone photos. Sending directly RAW to pano programs solve the problem but I lose the edits from lightroom. The noise reduction and removal color aberration in lightroom are excellent – Vic Sha Mar 8 '18 at 15:58
  • "I didn't edit the raw file, simply loaded it and exported it as Tiff 16bits." When you open a raw file in LR even if you don't edit anything, Lightroom certainly applies very significant processing to the raw data in the file. The TIFF you are saving out of LR has that processing applied to the raw data. – Michael C Mar 8 '18 at 16:47
  • You still seem to think that what you see on your screen when you open a raw file is "the raw data". It is not. It is one possible interpretation of the raw data among a near infinite number of possible interpretations. – Michael C Mar 8 '18 at 16:52
  • Michael, I agree with everything you say. But that doesn't change my point, that you confirm, that RAW to TIFF 16 bits loses color depth. What is sad about this is that there are not alternatives. We could convert raw directly to 32 floating point bits that would avoid this kind of problems but there is no single program that allows it. And this create this bottleneck where I can not send edited RAWS to HDRI pano programs without losing a couple of exposure steps. In general the most advanced 32 bit editing tools are ironically in composing programs leaving photography behind. – Vic Sha Mar 8 '18 at 19:43
-1

There's more information in the RAW than can be displayed on the screen. Thus, you can have a completely white spot in your current representation of the raw data, but lowering brightness will reveal some of that hidden data.

When you export to TIFF, PS will export the image as-is, i.e. complete white will be 65536 and complete black will be 0. So there's nothing to recover in this image anymore. To retain all information, you have to adjust the image, e.g. by lowering the contrast, so that all the image data falls between the two extremes.

With a TIFF like this, you will again be able to create any representation you could get from the original RAW, by increasing contrast again and manipulating brightness / exposure / highlights / whatever.

The advantage of the 16 bit vs. the 8 bit format is that you have a lot more of gradations between the extremes. If you tried the same with an 8 bit format, the "compression-decompression" step (by reducing and re-increasing contrast) would introduce horrible posterization since you'd loose a lot of the in-between colours.

  • From I can see you are right. RAW contains info that is beyond what is displayed, as 32 bit floating point but a lot less. Meaning that If a pixel near to the sun appears white 255-255-255 it can look a normal RGB with 3 numbers with different values when reducing the exp. When in TIFF 16bits simply adds more between pixels but anything not visible is lost. That is quite sad as there is no way to send edits from lightroom to other programs without losing data. I have more quality sending them directly the RAW but then I lose the edits. – Vic Sha Mar 8 '18 at 15:30
  • i have described how not to lose data: you have to take care that the image displays all shades from the darkest to the brightest. you can see in the histogram if anything is clipped. when you have an already edited image in LR which you want to process further, the missing data is not a problem since it already looks like you want it anyway. – ths Mar 8 '18 at 15:43
  • I use a secuence of bracketed 3 photos to create a HDRi 32 bit EXR panorama. What I want to avoid is to confuse the program that needs to create the HDRi toning (compressing) the bracketed images. It is something I will test anyway. Another alternative is creating an artificial extra bracketed image changing the exposure to -2 (and +2 for the higher exposed) and then creating a secuence of 5 bracketed images, 2 artificially created. This could avoid the lost exposure info. – Vic Sha Mar 8 '18 at 15:46
-1

But for my great shock I have noticed that there is a considerable lost of depth.

Make certain that your monitor is correctly calibrated. Can you distinguish every square?

Gray Chart

Since your question's example shows a different cropping than the files that you drop boxed I had to include the original .tiff, for equal comparison, so it didn't look like I used a different photo.

Original .tiff converted to .png and reduced 27%, the saved as .jpg

I downloaded the .tiff, converted it to .png, and used Pixlr Android to adjust the photo. The first (leftmost) image has been excessively adjusted, the remaining had the EV decreased only.

  • 1st - Used "Shadows", "increase 100 five times" "Brightness", "increase 10", "Contrast" "increase 10", "Saturation", "increase 10", and a few more ... Notice the brighter foreground grass, the slightly brighter Sun and the greater detail in the clouds. Looks like 16 Bits.

  • The 2nd, 3rd, 4th, and 5th are "Exposure", "decrease 50", "decrease 100", "decrease 150", and "decrease 200".

The .tiff is: RGB 16 Bit, each image was scaled to 25% after making the adjustments, then converted to a collage, then scaled down to 20%. After all that messing you'd expect a reduction in quality, it seems to me that the result is not so bad; do you have it set to reduce on import?

Collage made from Vic Sha's photos

I see no banding or anything to indicate loss of bit depth.

  • But at the end I lose nearly or more 2 exp, when the sun looks white in -3 in RAW it looks grey in TIFF. This affects for example the use of image for 3d lighting and the exp info is used for the light intensity. You can check also the longer post with extra images that I did copying your layout. – Vic Sha Mar 8 '18 at 15:17
  • @VicSha - My answer is not so different from MC's or ths's, the RAW is converted to display on your monitor so you can see a representation of it; it's like tone mapping from HDR to display it on a device with fewer bits. When you look at the RAW and adjust the sliders the output is mapped so it's not crushed; the .tiff isn't being mapped the same way. I was able to take your .tiff (can't work with RAW on my phone), make lots of adjustments, and still end up with a smooth image without banding; even from a baked in photo. Notice how I darkened the photo but the Sun remains visible. – Rob Mar 8 '18 at 17:27
  • I agree with your comments about RAW. But about your TIFF comment I need to agree with other posts instead. You seem to imply the data in the Tiff has not been degraded but the only thing you do is to stretch the tiny values of the 16 bits, that can give more detail that is apparently visible. But if a pixel is 000 in Tiff, that is, there is no return to that. Using exposure simply give you grey, and that is that happens in your edit when RAW sun remains white for a while longer than Tiff when lowering exposure. There is hidden data in RAW unavailable in Tiff. – Vic Sha Mar 8 '18 at 19:33
  • 1
    Thanks. It is. The tiff is not degraded and I can not make detail that didn't exist in the first place unless I draw it in. We've explained baked vs. not. Choose the answer you like. A few of these answers are different ways to explain exactly the same thing mapping / baked - you're not losing "bits or depth" you're losing the remapping of your wide-ranging adjustments to a file format that isn't remapped on-the-fly. It's like converting from HDR to LDR without losing bits then saying you can't tone map again; you could but you'll have lost a wider range of adjustability. – Rob Mar 8 '18 at 19:55
  • Well, I suppose we agree but consider the problem from different points of views. Tone mapping in order to keep that extra data invisible in RAW is not an option for me because I need to build a invisible range of 32 bits with bracketed photos. The sun must be 7 or 9 steps depth. Having the sun visible in a toned image it is not use for me as then it loses the depth. It won't work with 3d programs. – Vic Sha Mar 8 '18 at 20:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.