If I remember correctly, there was a discussion on the list related to the color rendering in low light and ISO invariance. I don't rember all the arguments, but I have two factors in mind that may affect the colors.

But first, - just in case: have you checked that both the temperature and the tint numbers (as seen in LR) are the same for both photos?

1. With the colors that have very low RGB numbers in one or two color channels, it is easy to imagine that small variations due to discretization play a noticable role in the outcome once amplified (i.e. amplification of the rounding error).

2. With the low levels in one or two color channels, you effectively compress the dynamic range at the lower end (which, I suspect might even yield bending, as someone on this thready has mentioned already). [I am too busy/lazy at the moment to think carefully about this second argument. Please, don't kick too hard... ;-) ]

Actually, one other question: can the specifics of what LR does to the RAW file (what they refer to as "process version") affect the color outcome. I.e., if one used the process from the earlier versions of LR (e.g. LR 1.x) and from the latest, could the outcome be different?

Igor




On Fri, Jan 27, 2017 at 5:54 PM, Gonz wrote:

Ok.  I've done some experiments to test this whole ISO invariance
subject, which some ascribe almost mythological healing powers to
underexposed images.  I'll reveal one pair of experimental images
after I have my methodology down.  Hence this post.  The results were
so surprising, that it made me question my methodology.

Here is what I did.

1. take an image with ISO 1600 properly exposed.  Use manual and set
shutter and aperture for image result that takes up most of the
histogram (avoid blowing highlights)
2. take a second image with same shutter and aperture but at ISO 100.
I.e. 4 stops underexposed.
3. Import into lightroom, compensate +4 exposure on the underexposed
ISO 100 image.

My lightroom has a limit of +4, hence the selection of 100 and 1600
for ISO values.

Images should look roughly the same if this methodology is right?  Are
the ISO values correct?  100 * 2^4 = 1600, or is this wrong?


--
PDML Pentax-Discuss Mail List
PDML@pdml.net
http://pdml.net/mailman/listinfo/pdml_pdml.net
to UNSUBSCRIBE from the PDML, please visit the link directly above and follow 
the directions.

Reply via email to