We know that the JPEG produced by in-camera RAW conversion is very close to that produced by using the RAW converter provided by the camera manufacturer, but how close is it in detail?
I'm interested in this as a general question, but for a specific data point I looked at my Canon 550D (firmware 1.0.9) and Digital Photo Professional (version 3.11.26.0). What I did:
- Take a photo in RAW + JPEG; I used standard picture style, auto white balance, no automatic lighting optimizer, no highlight tone protection.
- Load the RAW file into DPP and convert it to JPEG at quality 7 - quality 7 chosen as that makes GIMP's view of the JPEG quality parameter match the in-camera JPEG at 97.
- Compare the two JPEGs. While they're pretty close to each other, they're not identical. As well as some trivial differences in the metadata, it looks like the in-camera JPEG possibly has slightly more sharpening than the DPP converted one. This is particularly obvious around the white flecks towards the bottom centre of the photo.
If you want to see the files here, I've put them all on Dropbox, including a PNG of the difference between the two JPEGs.
This is potentially two very related questions:
- Is it possible to get DPP to produce identical output to the in-camera JPEG engine?
- Is it possible for other camera manufacturers / own brand RAW converters?
For avoidance of doubt, this is almost entirely a theoretical point; I'm well aware the differences are small enough they're not going to make a difference in practice.
Answer
In theory the camera manufacturer can make the processing exactly the same or completely different since they have full control over what goes into the camera and the raw processor.
It's both "common knowledge" and "common sense" that the in-camera an stand alone raw convertor from the same company will produce the same results, except for the converter version: software is easier to update than firmware, the stand alone converter may have a different version of the same raw processing algorithm than what you have in camera (this is especially true for older cameras that no longer have regular firmware updates like the 550D).
However, an opinion that is both "common knowledge" and "common sense" doesn't have to be true - this is especially true since those two processing systems have different target audiences (pro/serious hobbyist for the external processor and the full-auto crowd for the built in processing) and that some camera companies outsource their raw processor entirely (for example Nikon) so it's possible the two will use completely different algorithms.
There's really no way to know for a specific camera model except to test it.
No comments:
Post a Comment