Feature request: save as floating-point

(Graeme W. Gill) #202

Yep, that’s part of it. I have a set of slides from the sRGB website too though (overview and structure).

(Morgan Hardwood) #203

@Elle do what’s best for you, health comes first, but know that your input is valued.

I think what you’re looking for is here:

  1. Click on your profile avatar in the top-right corner,
  2. Click the gear icon, “Preferences”.
  3. See “Notifications”.

(Desmis) #204

I appreciate your skills and exchanges with you :slight_smile:


(Glenn Butcher) #205


@Elle, your body of “published” work is substantial, and I’ll continue to learn from it. But your posts have been equally informative, and I’ll miss them.

I’d then go on to endorse some sort of voice transcription solution, but if the state of the art is my phone’s voicemail transcription, we’re not there yet… :smile:

All the best,

(Alan Gibson) #206

@Elle: you have my sympathies, and my thanks for your substantial contributions. I have learned a great deal from you.

As and when you want, I hope you will continue to publish, eg at https://ninedegreesbelow.com/ . Articles and tutorials can be worked on at your own pace, without the social media pressures of “I must respond today”.

(Clinton Ingram) #207

You’ve certainly got me there. I was going to say I would be extremely interested in seeing anything that documented how that profile came to be, but I see below that you were referring to the draft version of the sRGB spec, which I referenced several times in my blog post. That draft spec, in fact, forms the basis of my belief that the sRGB whitepoint and color primaries are distinct from Rec 709, even if only very slightly so.

My argument is that while the chromaticities were the starting point for computing the matrixes, the numbers were adjusted in the final spec such that they no longer precisely agree with the starting Rec 709 values. Nowhere is this more evident than in the incorrectly-rounded (or shall we say nudged?) Z value of D65 given.

I posted an update to my blog post that summarizes both sides of the argument, and I included some handy visuals for understanding the inconsistency within the spec. If you see anything wrong in those explanations, let me know.

My argument is that this isn’t necessary, because if you view the chromaticities only as a point of historical reference, everything in the spec is already balanced and precise enough for its stated design goals. And let’s not forget that we’re talking about using the standard within the context of an ICC profile, where we lose any added precision as soon as we convert to 16-bit fixed-point format.

I’m so sorry to hear that, Elle. I hope you’re feeling better soon. And again, thank you for sharing your knowledge on your website and here.

(Graeme W. Gill) #208

Actually no - I was referring to this (You’ll have to unzip it to see the slides). I’ve put a copy of the HP/MS sRGB profile here.

(Clinton Ingram) #209

That’s very cool. Thanks for digging it up!