[Tig] RE: Why isn't HDR(-like)

Dan C. Tatut dtatut
Thu Mar 30 16:37:00 BST 2006

Thanks to oktobor for supporting the TIG
Floating point has accuracy errors in the extremes... the very nature of floating point in computers. However for the use of floating point in this industry, it has enough precision.

For the rest, I would remind to all of you this:

Bill Gates said once that 640Kb should be sufficient for every one... Where are we now?

Best, Dan

Dan Tatut
CHROME Imaging
105 Rue de Lyon
CH-1203 Geneva

Phone: +41 22 807 23 60
Fax: +41 22 807 23 70
Mobile: +41 78 659 11 04
WWW: http://www.chrome-imaging.com

-----Original Message-----
From: Digital Praxis [mailto:digital.praxis at virgin.net]
Sent: jeudi, 30. mars 2006 17:06
To: tig at tig.colorist.org
Subject: [Tig] RE: Why isn't HDR(-like) 

Thanks to oktobor for supporting the TIG
Has anyone been able to overcome the accuracy errors with floating point calculations when compared to integer calcualtions? Not sure they matter, just interested.

Also, in the real world the use of extended range calculations with non HDRI images tends to produce more then enough image processing flexibility, without the unnecessary overheads of larger than required data files.

Remember, 10bit LOG images have inbuilt extended range (as was mentioned earlied. by Bob), which when combined with over-range internal provessing, say 16bit, will give more than enough flexibility for 'most' situations. Those that don't work can always be done with a slightly different approach - there is never a 'can't do' situation with this type of workflow.

It's Bob's second way, which is to me the most practical.

In my business I don't want to spend money unnecessarily just to be able to cover stupid and rare workflows. I want to make money from the work I do, by keeping costs sensible, which at the moment means keeping storage and networking (bandwidth) requirements to the minimum, while not compromising final image quality.

Linear files, or even tv gamma files, have a load of redundant information due to their poor highlight sampling when compared to Log, with the result that it takes a 14bit linear (tv gamma) bit bepth to match the same granulatity of a Log file. That's a lot of unnecessary overhead for no benefit.

At the end of the day, I can see no quality difference between 10bit Log images processed correctly, and HDRI images and processing. But I can see a difference between 2K and 4K, if the projection medium can actually show the difference, and the capture format actually has the MTF to capture the difference...

It's good once in a while to remind ourselves that this is a creative industry aimed at producing a useable final image that joe public will pay to see in one form or another - not a scientific community striving to take everything to the Nth level.

It's like that old joke about an engineer and a creative, both arriving in hell at the same time.

The devil shows them their life in hell will take the form of standing at the end of a long room, with a beautiful, naked, woman at the other end.

The problem is that once every 100 years the distance will half, carrying on for ever without end.

The engineer cries, saying 'but I'll never get there!'

The creative smiles, saying 'I'll get close enough!'


Steve Shaw
Digital Praxis Ltd
+44 (0)7765 400 908
steve at digitalpraxis.net
skype: shaw.clan
msn: shaw.clan

Tig list - http://tig.oktobor.com/mailman/listinfo/tig
TIG wiki: http://tig.colorist.org/wiki

Tig list - http://tig.oktobor.com/mailman/listinfo/tig
TIG wiki: http://tig.colorist.org/wiki

More information about the Tig mailing list