Originally posted by K2 to K50 Not at all - would be very interested to see what you come up with, eaglem. I presume you mean to download the raw file on the OneDrive link? But whichever, go fo it!
I found that interesting too and hope that's an open invitation, even though it was directed at our resident rose expert. I downloaded the DNG and checked: It's well-exposed in the sense that no channel is clipping and noise in the darker parts still lets textures on the petals come though - just. Saturated reds are somewhat tricky to render in sRGB. The red primary color is fairly "pale", so the red channel easily clips. For an illustration: all the visible reds outside of the sRGB triangle (color not to scale) in the
chromaticity diagram have to be mapped into the triangle. The situation is not much better with AdobeRGB, btw. E.g., the color spaces of newer video standards cover a much wider gamut in the reds. Of course, we can't work with more than the sensor is able to differentiate either, so it is not as severe of a limitation as it looks.
Back to the RAW data. I tired a somewhat natural rendition to demonstrate what a largely non-clipping version would/could look like. Starting from a 'neutral' (fairly flat) rendition, I tamed down the highlights by flattening the tone curve in the higher values (Highlight compression = 67 in Rawtherapee). The camera's color balance measurement was off due to the amount of red. Being mostly backlit by sunlight, I manually set it to 'shade' (=6200K). I adjusted contrast to +10 (default in RT is very flat) and first lowered exposure, until only a tiny fraction of the petal highlights remained clipped. That turned out very dark, so I added a virtual graduated density filter of 1EV strength, gradient going from bright in the lower left to dark in the top right, dimming the highlights but still looking natural, which allowed to pull exposure back up to +/-0. A -0.6EV vignette helps to make the subject stand out a little more (dampen the right-top highlight more without affecting the center). Finally, I used RT's Tone Mapping (tamed down to 0.25 from way too big default of 0.5), resulting in increased local contrast and thus accentuated texture and stamens and lifting (too) deep shadows. The result looks OK to me for rendering on a monitor, preferably with a dark background (try the style selector at the bottom of the forum page). It would not print well at all though.
For a brighter background, like the default here in the forum, or for a print, which provides much lower contrast (~1:100) than the backlit screen (~1:1000), one has to trade off color fidelity for brightness. In order to partially compensate visually, as eaglem suggested, contrast can be increased. In the brighter example, I tweaked the tone curve for higher brightness and stronger contrast on the petal mid-tones, used RT's 'Local Contrast' settings (focusing on brighter parts) for even more accents and 'Dynamic Range Compression' (a global tone mapping operation) to fit everything into a smaller tone range. Even though saturation is clearly lacking in places now and colors are
slightly off (easy to totally screw up if
any of the above is pushed too far), this highly processed version still looks like a convincing rendition of a backlit red rose in bright daylight to me.
A white or silver reflector to brighten up the flower using natural light would likely have made the post-processing much easier.