Reading mode: Light Dark

Why is the Fuji RAW file bigger than Sony and Olympus?

Go to last post
Amnon G

Amnon G

Well-known member
Messages
232
Reaction score
21
Location
Bellevue, WA, US
Browsing through DPReview image quality comparison pages I noticed that with same resolution files, the Sony NEX-6, MEX-5N and Olympus OM-D EM-5 are about 14MB-16MB whereas Fuji X-E1 RAW files are about 25MB.

What's the reason for that?

Potential reasons (without knowing):
  1. More bits per pixel in RAW format (the Fuji would have to have 66% more bits, e.g. 12 vs 20 or 8 vs 13, which is obviously not the case)
  2. Better lossless compression (aren't all using pretty much the same lossless Lempel-Ziv compression?)
  3. Lossy compression (I remember at least the Nikon D90 threw away some RAW data compared to the more professional models, don't remember the details).
Anyone knows for sure?

Thanks,

Amnon
 
Amnon G said:
Amnon G wrote:

Browsing through DPReview image quality comparison pages I noticed that with same resolution files, the Sony NEX-6, MEX-5N and Olympus OM-D EM-5 are about 14MB-16MB whereas Fuji X-E1 RAW files are about 25MB.

What's the reason for that?

Potential reasons (without knowing):
  1. More bits per pixel in RAW format (the Fuji would have to have 66% more bits, e.g. 12 vs 20 or 8 vs 13, which is obviously not the case)
  2. Better lossless compression (aren't all using pretty much the same lossless Lempel-Ziv compression?)
  3. Lossy compression (I remember at least the Nikon D90 threw away some RAW data compared to the more professional models, don't remember the details).
Anyone knows for sure?
Click to expand...
I think it is 2) The Fuji files can be compressed, using something as simple as bzip2, by up to 80% (depending on the scene). Nikon losslessly compressed files can't be further compressed; they seem to have figures it out quite well.
 
I am not sure myself but dpreview say in their review of the X-Pro1 :

"There are several points to be made here. Firstly, the X-Pro1 shows visibly higher resolution in this chart test than a conventional 16MP Bayer-type camera such as the Nikon D7000, or even the Sony NEX-5N (which has a particularly weak AA filter). In fact, in terms of resolution it's very close indeed to the 24MP Sony NEX-7 or the Sigma SD1"...

I am not sure whether file sizes are in relation to this, I can't see how they can be, but there's no doubt the X-Trans sensor has superior apparent resolution than the cams you mention, and in controlled lab test dpreview have conducted also have found this.

Embedding stills for TVC work X-Trans images need the least prep than most other cams with the exception of the monster Canon and Nikon full frame cams. This might have something to do with it too.

Whatever the reason I am sure glad I have an X-E1, it has blown me away in every respect.


Cheers
 
BillyInya said:
BillyInya wrote:

I am not sure myself but dpreview say in their review of the X-Pro1 :

"There are several points to be made here. Firstly, the X-Pro1 shows visibly higher resolution in this chart test than a conventional 16MP Bayer-type camera such as the Nikon D7000, or even the Sony NEX-5N (which has a particularly weak AA filter). In fact, in terms of resolution it's very close indeed to the 24MP Sony NEX-7 or the Sigma SD1"...

I am not sure whether file sizes are in relation to this, I can't see how they can be, but there's no doubt the X-Trans sensor has superior apparent resolution than the cams you mention, and in controlled lab test dpreview have conducted also have found this.

Embedding stills for TVC work X-Trans images need the least prep than most other cams with the exception of the monster Canon and Nikon full frame cams. This might have something to do with it too.:-D
Click to expand...



Billy, you are hilarious. Someone asks about file sizes, something completely unrelated to IQ, but you still manage to put your usual shill stuff in your response.

BillyInya said:
Whatever the reason I am sure glad I have an X-E1, it has blown me away in every respect.
Click to expand...
I think your X-E1 hasn't blown you away far enough :-D :-D :-D
 
Well, here's a heads up for you, file sizes can indeed be associated with the resolution.

I don't really think there is anything hilarious about dpreview's findings. I know positive findings about the X-Trans sensor rarely sits right with you, but I happen to think the dpreview testing regime is very good.

And they do have a point when it comes to their findings on the apparent resolution of Fuji X-Trans sensor - which eclipses other 16MP sensor cameras with the old Bayer pattern.

Cheers
 
nixda said:
nixda wrote:

I think it is 2) The Fuji files can be compressed, using something as simple as bzip2, by up to 80% (depending on the scene). Nikon losslessly compressed files can't be further compressed; they seem to have figures it out quite well.
Click to expand...
I'll check that out. If I can at least compress them on my computer it would be great. To reply to others here - IQ has very little to do with the file size. That is debatable and would depend on compression schemes (losslessly-compressing red, green and blue as contiguous parts rather than interspersed, for example, may have an effect depending on chroma noise, for example), but not by this much.

I wish FUJI officials would look at these forums and give official responses, though... They might get some good ideas too :-)
 
BillyInya said:
BillyInya wrote:

I am not sure myself but dpreview say in their review of the X-Pro1 :

"There are several points to be made here. Firstly, the X-Pro1 shows visibly higher resolution in this chart test than a conventional 16MP Bayer-type camera such as the Nikon D7000, or even the Sony NEX-5N (which has a particularly weak AA filter). In fact, in terms of resolution it's very close indeed to the 24MP Sony NEX-7 or the Sigma SD1"...

I am not sure whether file sizes are in relation to this, I can't see how they can be, but there's no doubt the X-Trans sensor has superior apparent resolution than the cams you mention, and in controlled lab test dpreview have conducted also have found this.

Embedding stills for TVC work X-Trans images need the least prep than most other cams with the exception of the monster Canon and Nikon full frame cams. This might have something to do with it too.

Whatever the reason I am sure glad I have an X-E1, it has blown me away in every respect.

Cheers
Click to expand...
I think you're confusing perceived resolution with actual, pixel-based resolution. The former is subjective; the latter is not.
 
BillyInya said:
BillyInya wrote:

Well, here's a heads up for you, file sizes can indeed be associated with the resolution.
Click to expand...
True. But what amnon is talking about is due to compression I reckon. Less risk with less compression in my opinion which is good.
BillyInya said:
I don't really think there is anything hilarious about dpreview's findings. I know positive findings about the X-Trans sensor rarely sits right with you, but I happen to think the dpreview testing regime is very good.
Click to expand...
True. "halarious" is not a term I would use to describe what dpreview say in their reviews, even if I disagreed with it.
BillyInya said:
And they do have a point when it comes to their findings on the apparent resolution of Fuji X-Trans sensor - which eclipses other 16MP sensor cameras with the old Bayer pattern.
Click to expand...
I have had an X-Pro1 for a week now, purchased used. Pleased as punch. Color is insane, resolution is out of sight better than I got with my D300 or Nex5 and low light stuff is sensational.

Shots taken at IS01600 are like my Nikon and Sony at ISO400.
 
BillyInya said:
BillyInya wrote:

Well, here's a heads up for you, file sizes can indeed be associated with the resolution.

I don't really think there is anything hilarious about dpreview's findings. I know positive findings about the X-Trans sensor rarely sits right with you, but I happen to think the dpreview testing regime is very good.
Click to expand...



I enjoy my X-E1 as much as you seem to enjoy yours. I bought into the X-Trans, so don't think that I wouldn't want it to do well.

BillyInya said:
And they do have a point when it comes to their findings on the apparent resolution of Fuji X-Trans sensor - which eclipses other 16MP sensor cameras with the old Bayer pattern.
Click to expand...
File size, primarily has to do with the number of pixels and the bit depth. Add some amount of space for a thumbnail preview, the EXIF data and some other metadata, and you'll have your final size.

File size has nothing to do with the effective resolution that you are referring to. That resolution comes after processing the files, and RAW files are before processing them. JPEG files can have different sizes depending on the resolution (obviously), but we aren't talking about JPEG files.
 
Oz Rat said:
Oz Rat wrote:
Oz Rat said:
BillyInya wrote:
Click to expand...
Oz Rat said:
I don't really think there is anything hilarious about dpreview's findings. I know positive findings about the X-Trans sensor rarely sits right with you, but I happen to think the dpreview testing regime is very good.
Click to expand...
True. "halarious" is not a term I would use to describe what dpreview say in their reviews, even if I disagreed with it.
Click to expand...
I said that I find Billy hilarious (benevolently expressed), not the DPReview findings. "Hilarious" because of his endless shilling, no matter what the subject.
 
BillyInya said:
BillyInya wrote:
And they do have a point when it comes to their findings on the apparent resolution of Fuji X-Trans sensor - which eclipses other 16MP sensor cameras with the old Bayer pattern.
Cheers
Click to expand...
you kind of forget that it is not because of xtrans CFA, but because of no AA filter... take regular 16mp bayer w/o AA filter and XTrans will be put to the rest w/ subpar demosaicking... which will never ever be fixed...
 
nixda said:
nixda wrote:
nixda said:
Oz Rat wrote:
nixda said:
BillyInya wrote:

I don't really think there is anything hilarious about dpreview's findings. I know positive findings about the X-Trans sensor rarely sits right with you, but I happen to think the dpreview testing regime is very good.
Click to expand...
True. "halarious" is not a term I would use to describe what dpreview say in their reviews, even if I disagreed with it.
Click to expand...
I said that I find Billy hilarious (benevolently expressed), not the DPReview findings. "Hilarious" because of his endless shilling, no matter what the subject.
Click to expand...
Sorry, misunderstood.

So you do not have an issue with dpreview finding the x-Tans sensor to have visibly higher resolution more like a 24MP Bayer sensor?
 
to be fair, though, if the sensor does capture more high-frequency detail - its output should be less compressible than that of a sensor that captures blurry images, at the same pixel count.

Not saying this is necessarily the case with Xtrans (mostly likely Fuji just can't be bothered to optimize raw compression), but still.
 
Imagefoundry said:
Imagefoundry wrote:

to be fair, though, if the sensor does capture more high-frequency detail - its output should be less compressible than that of a sensor that captures blurry images, at the same pixel count.

Not saying this is necessarily the case with Xtrans (mostly likely Fuji just can't be bothered to optimize raw compression), but still.
Click to expand...
The Fuji RAF files from the X-E1 are all 26146816 bytes large. They are not compressed, or Fuji found a magical way to always come up with the same file size even after compression. Unlikely.

There have been threads before about the file sizes. Some cameras write out larger files than others, regardless of resolution. Cameras can record 12 bit data but write out 14 or even 16 bits.

To link the file size with the resolution of the image is, IMHO, quite far-fetched.

I have taken images with my D7000 and with my X-E1. We know that the D7000 images are compressed (lossless). After compressing the X-E1 image, the file size was smaller than that of the D7000, but I certainly wouldn't go around now claiming that the D7000 image has a higher resolution.
 
to my chagrin, I didn't even know that Fuji RAFs are not compressed... just never bothered to look... (this seems pretty archaic!). Yep, each one is 25,534KB... My bad.

My comment was rather about detail in the capture affecting compression ratio, much the same as a jpeg of a out-of-focus picture is invariably smaller than the one of in-focus picture - but it's pretty irrelevant in the context.
 
The DPR statement is not reliable! The resolution number is never mentioned and DPR clearly states it in every other review I have ever seen. Maybe they lost track of the real data in the barrage of all the false detail in the X-trans chart. Personally I like the false detail.

I took a look and compared the resolution chart with the D7000. Except for the moire they look about the same to me. If you try this, count the number of stripes. When the number changes the sensor resolution is exceeded. Not difficult but you do have to pay close attention.

Interesting that Popular Photography says the X-trans diagonal resolution is better than the vertical and horizontal resolution. Not possible to evaluate this by looking at the DPR resolution test shot. DPR should make a few improvements I think.

I also own a Sigma SD1 (15 megapixel Merrill sensor). The X-E1 resolution is not in the same league, but it's still pretty good. The X-E1 is the camera I carry, not the SD1. I tried last week to clean the SD1 sensor, but that's a whole different thread, in a whole different forum!

The X-E1 supersonic sensor shaker runs regularly whenever I shut down the camera while in shooting mode. Great feature!

Back to the raw file size: there is no time taken to run any sort of compression algorithm, so all that is necessary is to write the data. Sounds pretty simple and efficient to me. The camera might actually run a little faster in bursts this way.

I was thinking about setting my X-E1 to store Raw+JPG, and now I see the data loading would go from 5 meg or so for JPGs only to 30 meg or so, per shot.
 
Last edited:
By the way, the idea of compressing later, on the computer, is pure genius! Yet another reason to shoot Raw+JPG all the time: The Raws are only big in the camera card!
 
nixda said:
nixda wrote:

The Fuji RAF files from the X-E1 are all 26146816 bytes large. They are not compressed, or Fuji found a magical way to always come up with the same file size even after compression. Unlikely.
Click to expand...
23970816 bytes of raw data (12bpc), if it were 14bpc the raw data alone would be 27965952 bytes. 65450 bytes of EXIF (seems to be a fixed size). That leaves around 2110550 bytes unaccounted for. That makes uncompressed very likely.


That's also supported by the easy 20-30% compression available on these files, vs <5% on an OM-D raw file of the same scene.

Plus dcraw uses unpacked_load_raw to handle RAFs, so that sort of settles it. Reading the source would answer most of these questions definitively, if one has the time.
nixda said:
To link the file size with the resolution of the image is, IMHO, quite far-fetched.
Click to expand...
It displays a complete lack of understanding of the topic under discussion. Always a great opportunity for some mindless copy/pasting though. :)
 
Tom Schum said:
Tom Schum wrote:

By the way, the idea of compressing later, on the computer, is pure genius! Yet another reason to shoot Raw+JPG all the time: The Raws are only big in the camera card!
Click to expand...
 
It appears that the Fuji raw files are not compressed. Every raw file that I have shot is exactly the same size (26,246,816 bytes). I even took a picture with my lens cap on and got exactly the same size (if any file could be successfully compressed, one that is total black should be the one). By the way, all of my shots are in 3:2 aspect ratio. The pixel count of my images is 4896x3264 or 15,9980,544 pixels. I don't know how much overhead there is in the raw file, but I thought I would see what the maximum bit depth of the raw file could be. If you assume a bit depth of 12, the minimum space you could store that data would be 23,970,248 so that could fit. At a bit depth of 13, the minimum space would be 25,968,384 which could also fit. At a bit depth of 14, the space required is 27,965,952 so that will not fit.

I have not seen anyplace that Fuji specifies the bit depth for raw files, so, based on the numbers above, I am guessing 12.

And as to the original question, I think it is that the raw files are not compressed whereas the other manufacturers are compressing the raw files.
 
You must log in or register to reply here.

Similar threads

A
Olympus OM-d E-M10 Mark III vs Fujifilm X-T20 vs Sony Alpha A6300
2 3
Replies
45
Views
317
RSSrsvp
R
R
  • For sale For sale
Best Infrared & Full Spectrum converted cameras for $399 Only! Olympus E-M5, Sony NEX-5 Fuji X-E1
Replies
2
Views
3
Rensol
R
S
Fuji x-E1 vs. Sony A6000 vs. Olympus OM-D
Replies
13
Views
14
kwa_photo
kwa_photo
J
decision: Sony A6000, Olympus OM-D E-M5 II or Samsung NX500
Replies
3
Views
4
photosen
P
paul spicer
Why I purchased an Olympus OM-D E-M1
Replies
9
Views
10
DonParrot
D

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top