Question about profiles

look for ready-made device noise profiles and offer your own ones here
Post Reply
janarne
Posts: 27
Joined: Sun Aug 02, 2009 10:15 pm
Location: Saronida, Greece

Question about profiles

Post by janarne »

Hello i am actually making a set for the Olympus e-420 which isn't allready fully listed (only raw) i have aquestion to better understand the profiling:

1. i make a profile with a target for example,
the auto profiler chooses a place in the right upper corner on a bright square, and i save this the filesize is 38KB or sometimes 40KB (quality 94%)
2. now i just move the selector to a darker place at the same target, and re-scann it, and saved it as a new Profile (quality also 94%)
the filesize is 42 KB.
questions:
1.where is the difference exept the filesize and why?
2.are bigger or smaller filesizes "better" or dont care?
greets from greece
NITeam
Posts: 3173
Joined: Sat Feb 01, 2003 4:43 pm
Contact:

Re: Question about profiles

Post by NITeam »

jani wrote:1. i make a profile with a target for example,
the auto profiler chooses a place in the right upper corner on a bright square
Do you use Auto Profile with Regular Image or Auto Profile with Calibration Target (please see in the Profile menu)?
jani wrote:1.where is the difference exept the filesize and why?
2.are bigger or smaller filesizes "better" or dont care?
The difference is in the noise sample you used for main analysis.
The file size is not important. What is important with the Calibration Target is to use the correct square for main analysis. When you use Auto Profile with Calibration Target, it usually automatically selects the right square.

Vlad
janarne
Posts: 27
Joined: Sun Aug 02, 2009 10:15 pm
Location: Saronida, Greece

Re: Question about profiles

Post by janarne »

NITeam wrote:Do you use Auto Profile with Regular Image or Auto Profile with Calibration Target (please see in the Profile menu)?
In this example i mean i used the Caliration Target.
But i use also regular images if they produce high quality Profiles.
That happens for example in images with low deph field sharpness, where are sharp and unsharp parts in the image.
NITeam wrote: The difference is in the noise sample you used for main analysis.
The file size is not important. What is important with the Calibration Target is to use the correct square for main analysis. When you use Auto Profile with Calibration Target, it usually automatically selects the right square.
Do you mean with "right square" , the square with the best profile quality?
Yes i use first the Auto Profile function, and then i check 2-5 other places in the Calibration Target if there are better results. Sometimes i get better results when i check the squares by hand (at about 20% of all Calibration Targets)
greets from greece
NITeam
Posts: 3173
Joined: Sat Feb 01, 2003 4:43 pm
Contact:

Post by NITeam »

Please open the Profile menu and find two different types of Auto Profile there: Auto Profile with Regular Image or Auto Profile with Calibration Target. I was referring to those two different functions.

Vlad
janarne
Posts: 27
Joined: Sun Aug 02, 2009 10:15 pm
Location: Saronida, Greece

Post by janarne »

oh, i found it i used always the function "auto profile with regular image",
when i made the profile with calibration targets, is that very bad?
greets from greece
NITeam
Posts: 3173
Joined: Sat Feb 01, 2003 4:43 pm
Contact:

Post by NITeam »

Not very, but could be much better if Auto Profile with Calibration Target was used as suggested by the procedure described in the user guide. My recommendation - please closely follow that procedure when building profile sets.

Thank you,
Vlad
janarne
Posts: 27
Joined: Sun Aug 02, 2009 10:15 pm
Location: Saronida, Greece

Post by janarne »

I have readed the documentation before, but had forgotten that detail.
In the documentation is written a profile quality higher than 75% is good,
i aim a quality of 91%-100% (usually 93-96%), so thats no the problem i used "auto profile with regular image" on calibration targets.
In the future i remember to use "auto profile with calibration target".
greets from greece
janarne
Posts: 27
Joined: Sun Aug 02, 2009 10:15 pm
Location: Saronida, Greece

Post by janarne »

i have calculatet that to make profile for every thinkable set i could produce about 29700 Profiles, because each factor multiplies the possibilities.
For example
5 different Iso settings,
about 33 shutterspeeds (under 1sec),
9 White-balance Settings,
5 Image Sizes
4 De-Noise-Filer settings

offcourse thats too much, even for me ;-)
but it is theoretical a fact.:idea:

I have another question which i didn't find in the user guide,
how many Profiles are needed or good for a profile set?
and how many for a very good profile set?

usually it varies by 15-60 files i saw by the profiles i downloaded.

So is it possible when i make a set of about 290-350 profiles to publish it?
(offcourse no dubs)
Where is the upper maximum (2000?)
greets from greece
NITeam
Posts: 3173
Joined: Sat Feb 01, 2003 4:43 pm
Contact:

Post by NITeam »

The user guide contains a table that includes several noise-related parameters listed in the order of their significance. The user guide suggests to select some of those parameters starting from the top and build profiles for each combination. When selecting profiles to be included in this variable set you can decide how many profiles overall you can build and which combinations are really important for you, which ones you actually use. There is absolutely no need to profile all possible combinations.

For some examples of good profile sets please check the online library of profiles on Neat Image website. All of those profiles were built by Neat Image users who followed the recommendations of the user guide.

Thank you,
Vlad
Post Reply