"lossless compressed" on Fuji X T-2?

[Eric Brody]

I love my new Fuji X T-2 and have looked in some detail at a variety of images, with both shadow and highlight detail. I am unable to tell the difference between those made without compression and those made with "lossless compressed." The manual refers to a "reversible algorithm," with no loss of data. While storage is inexpensive, I still would like to conserve space, and presumably speed up processing in Lightroom and Photoshop. But I will not tolerate quality loss. In my Nikon D800E era, I always used uncompressed RAW files, possibly just from stubbornness and a belief (not founded on expertise of any kind) that there's always a cost to compression. I used the "if it sounds too good to be true..." argument.

Has anyone who might have more expertise than I do in terms of image quality looked at this issue? I suspect I am not the only one with this question.

[Alan Smallbone]

I have not been able to see the difference, but then I am not expert. I wish Capture One would support the lossless compression, their new release supports the X-T2 now but only with non compressed images.

Alan

[rdonson]

I'm no expert but I don't see a quality difference from the lossless compressed RAW files.

[john beardsworth]

They do say "lossless" and one can imagine the uproar if it did prove to be lossy. I don't see any difference in Lightroom and I'm am happy to use the XT2's lossless compressed option (as I do with my D800).

Intuitively at least, there must be a price for the compression - but that need not be in the image data. It may be in processing time, for instance. So conceivably each individual file will take longer for the camera to save to the card, the buffer might fill up more quickly (only an issue if you use high frame rates), or more power might be drawn from the batteries etc.

John

[Telecaster]

If you've ever used a ZIP or StuffIt utility to bundle files together you've used lossless compression. Given that you can thus compress application as well as data files, there's no reason for concern when it comes to RAWs. I can't imagine any camera company would risk using a lossy compression scheme while claiming it to be lossless. Too much potential negative economic and reputational impact.

-Dave-

[pcgpcg]

The manual refers to a "reversible algorithm," with no loss of data....there's always a cost to compression. I used the "if it sounds too good to be true..." argument.

As someone else has already mentioned, there may be a cost, but it is not a loss of information or image quality. If there is any cost it is related to the extra processing required to compress and decompress the file. Lossless compression is commonly used (ZIP files as someone else has mentioned) in the data processing field and it refers to compression that is truly completely reversible. There are lots of algorithmic opportunities for compressing data in an image without losing information. For example, the most obvious is a large area of the same color. Instead of listing the color for every pixel in that area, you can just define the boundaries of the area and the color, which provides the same information, but requires far less bits to record the same information.

[af_ahoy]

Compressed is almost certainly faster – the minor speed penalty from compressing the data is more than offset by having to write less to the card – storage is massively slower than working with data in-memory; in fact it's fast enough that there's even benefits to using compression for your computer's RAM.

To put this into perspective, if things were scaled up and it took an hour to compress 1KB of data, it would take the equivalent of about 24 hours to write 1KB to the card.

[Eric Brody]

Many thanks to all who replied. It all makes sense now. I'll save time and storage and use lossless compressed.

[ErikKaffehr]

Hi,

Most vendors use some version of"Huffman coding". Huffman coding is lossless, so the uncompressed image can be reconstructed bit for bit.

This is quite different from a compression scheme like the one used by Sony, which used a tone curve type of approximation taking shot noise into account. With such coding the image can not be reconstructed bit for bit, but no real information would be lost as all information deposed of would be masked by shot noise. Shot noise is the natural distribution of photons reaching the pixels.

To that Sony added a "delta type" compression which saved min and max values for 16 pixels and represented with just 7 bit coding. That compression could cause artefacts over edges having very high contrast and that has been demonstrated in some cases.

So you can have reversible compression, like Huffman. Or virtually lossless like the Sony tonal curve or a compression with artefacts like the Sony delta coding.

I guess that if Fuji talks about lossless compression it is "Huffman coding". A more efficient coding is LZW, but I don't think it is commonly used in camera raw files.

Best regards
Erik

I love my new Fuji X T-2 and have looked in some detail at a variety of images, with both shadow and highlight detail. I am unable to tell the difference between those made without compression and those made with "lossless compressed." The manual refers to a "reversible algorithm," with no loss of data. While storage is inexpensive, I still would like to conserve space, and presumably speed up processing in Lightroom and Photoshop. But I will not tolerate quality loss. In my Nikon D800E era, I always used uncompressed RAW files, possibly just from stubbornness and a belief (not founded on expertise of any kind) that there's always a cost to compression. I used the "if it sounds too good to be true..." argument.

Has anyone who might have more expertise than I do in terms of image quality looked at this issue? I suspect I am not the only one with this question.