Details about RAW format

[Moglex]

:waiting:


:banghead: My head.

And that's just considering B&W with no gamma.

Add those two into the equations and things start getting really interesting.

 

[Garbz]

Moglex this conversation is enough to turn a man to drinking. :cheers:

 

[Moglex]

Moglex this conversation is enough to turn a man to drinking. :cheers:

Oh, good.

Let's keep it up then.

 

[Stosh]

Moglex this conversation is enough to turn a man to drinking. :cheers:

Yes please!

Moglex: at the risk of beating this thing to death, could you explain one thing? You agreed with Garbz that the JPG doesn't have to be a subset of the RAW. But earlier you said

My impression, from making tome mapped HDR's from a single exposure, is that the sensor's dynamic range in higher than that which would give a natural looking end result on a screen or printer and thus when the RAW is processed it uses only part of the range of values potentially present.
When you 'manually' process RAW you can move this sub-range which gives you the 'latitude'.

If the JPG and RAW have the same dynamic range of the original scene, then the original scene must have had a "normal" dynamic range. Either that or your JPG doesn't look normal (because it's trying to display too wide of a range). Is that what you're saying? I would agree with that which is completely based on manually processing images from RAW with Canon's proprietary software. As you adjust the brightness slider up and down, you can either reveal highlights that were previously blown out, or shadows that were previosly black providing your original image has a large enough dynamic range.

 

[Moglex]

If the JPG and RAW have the same dynamic range of the original scene,

They will only have the same dynamic range as the original scene if the original scene has a range that is equal to or less that the dynamic range of the sensor.

The RAW will have a rannge that is the lesser of the scene's dynamic range and the sensor's dynamic range.

If the sensor's dynamic range is greater than that which will appear 'normal' when displayed, and the scene is such that it makes use of that dynamic range, then the JPG will have a smaller range selected from a subset of the RAW image.

If you 'develop' the RAW image using different subsets of the contained range and combine the developed images using appropriate software the software will create, internally to the computer, an HDR image which can then be tone mapped and displayed.

then the original scene must have had a "normal" dynamic range. Either that or your JPG doesn't look normal (because it's trying to display too wide of a range). Is that what you're saying?

No, because either the camera or PC based softare will select a subset of the RAW data so that it presents a 'normal' looking picture.

As you adjust the brightness slider up and down, you can either reveal highlights that were previously blown out, or shadows that were previosly black providing your original image has a large enough dynamic range.

Exactly.

And if you make those different images which, as you have just demonstrated for yourself, contain, overall, a greater range than any single image, and combine them appropriately, voila! a tone mapped HDR.

 

[Garbz]

And if you make those different images which, as you have just demonstrated for yourself, contain, overall, a greater range than any single image, and combine them appropriately, voila! a tone mapped HDR.

Can I just check something with this? Take the image, open it in Lightroom, crank recovery to 100, crank fill light up to bring up the shadows (ok image now looks like arse due to lack of contrast) but does this resulting JPEG now match the dynamic range of the RAW, since this is effectively dynamic range compression?

 

[Moglex]

And if you make those different images which, as you have just demonstrated for yourself, contain, overall, a greater range than any single image, and combine them appropriately, voila! a tone mapped HDR.

Can I just check something with this? Take the image, open it in Lightroom, crank recovery to 100, crank fill light up to bring up the shadows (ok image now looks like arse due to lack of contrast) but does this resulting JPEG now match the dynamic range of the RAW, since this is effectively dynamic range compression?

I don't use Lightroom but I think the answer is that you don't need to do anything to the JPG to get it up to the same range as the RAW if it was originally mapped for maximum range and you extracted both ends of the RAW range when creating the JPG source components.

Let's make up a hypothetical world where the camera has a dynamic range of 10 bands and people are used to looking at images with a range of 6 bands. (The usual caveat that we're pretending everything is linear to keep it simple).

Now, when the RAW is developed it will select 6 contiguous bands and map those to the jpg.

By default these bands will be the central 6 and there may be detail free shadows and highlights.

If we use exposure compensation we can force the selection of either lower or higher sets of six bands and thus we can create a pair of JPG's with, between them, all 10 bands that the camera recorded.

When these two JPG's are imported into suitable 'HDR' software it will find the 'common ground', normalise the JPG's so this common ground is the same and, internally, create an image that now has all ten bands that the camera recorded. It can then tone map that image and display the result.

 

[Stosh]

And if you make those different images which, as you have just demonstrated for yourself, contain, overall, a greater range than any single image, and combine them appropriately, voila! a tone mapped HDR.

Can I just check something with this? Take the image, open it in Lightroom, crank recovery to 100, crank fill light up to bring up the shadows (ok image now looks like arse due to lack of contrast) but does this resulting JPEG now match the dynamic range of the RAW, since this is effectively dynamic range compression?

I don't use Lightroom but I think the answer is that you don't need to do anything to the JPG to get it up to the same range as the RAW if it was originally mapped for maximum range and you extracted both ends of the RAW range when creating the JPG source components.

Let's make up a hypothetical world where the camera has a dynamic range of 10 bands and people are used to looking at images with a range of 6 bands. (The usual caveat that we're pretending everything is linear to keep it simple).

Now, when the RAW is developed it will select 6 contiguous bands and map those to the jpg.

By default these bands will be the central 6 and there may be detail free shadows and highlights.

If we use exposure compensation we can force the selection of either lower or higher sets of six bands and thus we can create a pair of JPG's with, between them, all 10 bands that the camera recorded.

When these two JPG's are imported into suitable 'HDR' software it will find the 'common ground', normalise the JPG's so this common ground is the same and, internally, create an image that now has all ten bands that the camera recorded. It can then tone map that image and display the result.

This is seriously what I've been trying to say all along, except that my grasp of my native language seems to be lacking!:cheers:

 

[Garbz]

So the fact that the JPEG doesn't normally cover the entire RAW area is left up to the camera manufacturer / RAW converter programmer then? Or is there an aesthetic reason that the JPEG is "crippled"?

I have a theory that things like Nikon's D-Lighting camera settings do exactly this, cram all the RAW data into the JPEG right out of the box.

 

[Moglex]

So the fact that the JPEG doesn't normally cover the entire RAW area is left up to the camera manufacturer / RAW converter programmer then?

Yes.

Or is there an aesthetic reason that the JPEG is "crippled"?

If the sensor does indeed have a dynamic range that is greater than that which can be displayed then the reason to 'cripple' the JPG is that if you didn't you would be forcing a tone mapped HDR onto people.

I have a theory that things like Nikon's D-Lighting camera settings do exactly this, cram all the RAW data into the JPEG right out of the box.

I don't use Nikon so I don't know but it has always surprised me that there is no way in the Canon raw processing program to use the whole range of the image and tone map it.