Giant files after develop CR2

[unitizer]

Title is the whole problem. I am opening and developing CR2 files, then saving as AFPHOTO. The CR2 files are already large at 20MB, but the resulting AFPHOTO files are whoppers at 150MB each! That's right, one sixth of a gig per photo. I must have something set wrong but I rarely use RAW / CR2 so I'm not sure what. Exported JPEGs are fine at 1MB.

[walt.farrell]

One question: What output format are you Developing to? If you look at the Develop Assistant (View > Assistant Manager), there's a setting where you can choose 32-bit or 16-bit for your output format. Developing to 16-bit should give smaller files (but probably still large).

[unitizer]

It was set to 16 bit. Why would the resulting file be larger than the CR2 source? I mean doesn't "Develop" bake in the settings that are selected from the original RAW file? I can't imagine how a 7x size increase could result under any conditions.

[walt.farrell]

14 minutes ago, unitizer said:

Why would the resulting file be larger than the CR2 source?

Because the data formats are very different. You may fid this Affinity Spotlight article useful: https://affinityspotlight.com/article/raw-actually/

[unitizer]

Thank you, I'm familiar with RAW in general though. I've been shooting Canon RAW for years and also owned a RED camera. Maybe I am incorrect in thinking that the Affinity "Develop" step is supposed to mimic the "Open" step in Adobe's Adobe Camera Raw (ACR). With ACR, the adjustable parameters in the RAW file get baked into a PS file that can then be further edited, all when you press "Open". The resulting PS file size though is nowhere close to 7x, in fact IIRC it is roughly the same size or smaller. That makes sense -- once you bake the RAW you do not need the massive amount of data, and if you do it still resides in the original RAW/CR2 file. It is hard for me to imagine any instance where having such a massive file -- should I repeat it, it's 150MB -- would be useful. Something is just not right here. The document you suggest is not clear on what should be expected in a Develop step.

The solution is pretty easy I guess. Bake the RAW using Develop, then export to an intermediate format, like a juicy PNG or JPEG. Then bring it back into Affinity for further processing. Seems silly but the file size is just not negotiable for me.

[kirkt]

A 16bit per channel RGB file is that size. It has nothing to do with Affinity Photo. For example:

A 6000x4000 pixel RGB image at 16 bits per channel =

6000 x 4000 pixels x 3 channels x 16 bits per channel / 8 bits per byte = 144 MB. 
 

Use 8 bit or save as a 16 bit TIFF with compression.   Your raw files converted to RGB images at 16bpc have always been this big, until you save them as jpeg. Then they get automatically converted to 8bpc (jpeg is not a 16 bpc file type) and then compressed. 

Raw files are not images, they are a matrix of digital numbers in a pattern that gets processed into a full three-channel RGB image through a process called demosaicing.  Each “pixel” in the raw file has one “channel” and is often encoded in 14 bits:

6000x4000 x 14 bits per pixel / 8 bits per byte = 42MB 

The raw file may actually be smaller than that if you use compressed raw, for example.

 

kirk

[unitizer]

OK so I am definitely wrong about what "Develop" does. It sound like rather than baking the settings created in the initial opening of the CR2, it instead expands the source file fully into an AFPHOTO format. I guess there is a plus and a minus to that. The minus I have already covered, but the plus would be that the resulting file would have full latitude / depth for all further processing. My understanding in PS was that once "Open" was pressed, the full latitude of the RAW file was lost, in fact that is why only certain operations occur prior to "Open" (or Develop). I'm thinking now that the pre - Open/Develop changes are only those that are encapsulated non-destructively in the CR2 file. I will say that in my experience, operations done post-Open in PS are not nearly as detailed as those done pre-Open in ACR. Even comparing something simple like exposure in ACR versus an exposure layer in PS, they don't behave the same. I have not used RAW in Affinity enough to say if the same is true (obviously).

Thank you!

[Jowday]

This is an 12MP RAW file opened and saved in Photoshop and Photo as 16-bit then saved directly:

Pretty close. So the amount of image data saved is kind of close. That is one thing to investigate. But no drama here.

What is more interesting is how big the files grow when you add all types of layers and remove some? I thin Serif removes some data only when the amount of slack (unused data) in the file crosses some threshold for performance reasons. I am curious what this means when you work on huge projects. 

 

[Jowday]

7 minutes ago, unitizer said:

OK so I am definitely wrong about what "Develop" does. It sound like rather than baking the settings created in the initial opening of the CR2, it instead expands the source file fully into an AFPHOTO format. I guess there is a plus and a minus to that. The minus I have already covered, but the plus would be that the resulting file would have full latitude / depth for all further processing. My understanding in PS was that once "Open" was pressed, the full latitude of the RAW file was lost, in fact that is why only certain operations occur prior to "Open" (or Develop). I'm thinking now that the pre - Open/Develop changes are only those that are encapsulated non-destructively in the CR2 file. I will say that in my experience, operations done post-Open in PS are not nearly as detailed as those done pre-Open in ACR. Even comparing something simple like exposure in ACR versus an exposure layer in PS, they don't behave the same. I have not used RAW in Affinity enough to say if the same is true (obviously).

Thank you!

Develop only develops the image in Affinity with the chosen settings - and then you can save it as a Photo file or export it as a JPG/whatever file. For some incredible reason I will never understand Serif decided not to save metadata with the file - in a sidecar file that remembers the chosen settings. They never did - not in their old Photo editor PhotoPlus either. You can save some settings individually on each develop tab.

[unitizer]

In that case I don't understand why there is a "Develop" step. All that step does is create an AFPHOTO file based on the CR2 -- what else would a user be doing when they open the CR2? I mean at least ACR has a button to save as a JPEG, skipping the larger PS altogether. ACR is really a standalone application if all one is doing is adjusting basic RAW image data like exposure, WB etc.

[kirkt]

AP is not raw converter. It is a pixel editor. The Develop module makes the raw file into an RGB file you can edit. Then export it as a JPEG. No need to save the .aphoto file. 
 

Saving a JPEG from ACR just skips the opening step. 
 

kirk

[Jowday]

7 minutes ago, kirkt said:

AP is not raw converter. It is a pixel editor. The Develop module makes the raw file into an RGB file you can edit. Then export it as a JPEG. No need to save the .aphoto file. 
Saving a JPEG from ACR just skips the opening step. 

It has a RAW develop module that converts RAW data to pixels with all the features dedicated RAW converters has. So ...

[Jowday]

29 minutes ago, unitizer said:

In that case I don't understand why there is a "Develop" step. All that step does is create an AFPHOTO file based on the CR2 -- what else would a user be doing when they open the CR2? I mean at least ACR has a button to save as a JPEG, skipping the larger PS altogether. ACR is really a standalone application if all one is doing is adjusting basic RAW image data like exposure, WB etc.

You never edit anything in a RAW. You adjust parameters for the demosaic and adjustment proces that results in a directly saved file or an image open in Photo. Photo doesn't save the parameters. Other programs does.

Based on? You need the development of the sensor data to work with the output. Every RAW converter does this a little different. This phase is not converting Dollar to Euro, it is more like an interpretation of the sensor data. Capture One renders sensor data more organic like Olympus or Fuji Cameras, Adobes rendering is more digital and pixel oriented. You do not get the same rendering of the same RAW file from Adobe, DxO Photolab or Capture One. Thats why I own all three.

 

[Jowday]

This is what is stored in the RAW file together with tons of metadata and more:

https://en.wikipedia.org/wiki/Demosaicing