I just had a go with Filmic V6. It worked better when I reduced the latitude slider value.
I have attached the xmp file but it has been renamed sunflower.
Set white balance to daylight
Set the exposure at +1.0 EV
Move the latitude slider all the way to the left to get rid of the reddish tint in the flower.
Apply denoise profiled default
Apply sharpness aa filter preset from diffuse or sharpen module
Apply basic colorfullness preset from the color balance rgb module
Use the auto tune levels in filmic. This returns the red tint to parts of the petals. Manually adjust white relative exposure slider to the right to return the yellow colour for the petals.
Apply default local contrast
Apply default shadow and highlights
Apply the sharpness module at default values
Use the color zones module to increase the saturation of the blue sky
Create a new instance of the color zones module to lighten the yellow flower
Yes the color zones module is a favorite of mine, but I don’t sue it now as it is supposedly not a scene referred module. AP is creating a linear RGB version of it in his fork, and I hope the main DT versions gets it too.
I have heard AP’s comments on it, but since it comes later in the pipeline and the results are fast and good I will continue using it. I do hope AP creates an even better version of the the tool and I also hope it makes it into the main branch of DT.
BTW, I also reposted my Filmic V6 version after setting white balance to daylight rather than as shot.
not sure what happened, but the sunflower.cr2.xmp file looks incomplete when I load it and the result is different than the pic. But anyway you can see that the the v6 render still has that “flat” look, even though it does seem better than what I posted originally. One thing for sure is that unless you have something to compare against, it’s hard to asses how good it is. That is one reason I always use the SOOC version as a ref. It’s not to make may render look like the the sooc, far from it. Rather it helps me know if I’ve improved it or not.
I often do the same with the camera JPG. I can look at it and say I prefer the contrast/color or something else from the JPG and do something about it with the RAW.
Not sure what the problem with the XMP file was, but I am using the latest weekly build for windows so it may be a backwards compatibility issue.
BTW, I was impressed with your original edit and the approach you took was very involved. I tried to simplify it by using the color zones module instead of masks.
Why bother about an artistically effect is done scene referred or not. That boost to the sky isn’t natural at all, so it’s no longer about preserving the scene…
You can use base curve instead of filmic no problem if that’s what you want. It makes giving things a nice highlight rolloff a bit harder IMHO , and you have to work way more with tone equalizer before it to get your tones where you want them (and people have often said that module is also hard to work properly with te masking and stuff ). But for images where you want something like the SOOC punch and you don’t have lots of dynamic range in your scene it might actually be a quick way of working.
I generally just use filmic on everything , but every once in a while you get a problematic image , And this is one of the things you could try.
If you push a slider a little bit, it is technical, but a bit more and it’s artistic. I don’t care about distinctions like this so much. I invented a game where I win more whimsy dollars if I can get what I want and stay within the scene referred portion of the pipeline. Points are taken away though for overcooking the pasta.
So since brilliance tries to preserve chrominance, I found that with v6 i can reduce the brilliance of the highlights some degree using color balance rgb, and it delays the onset of the color breakup when increasing the overall intensity. But it still doesn’t get as far as I’d like, and if you try to do too much it just takes on that flattish look again.
Here is my version, personally I prefer to use the V6 of filmic in synergy with the RGB colour balance and colour calibration module. I opted for more dynamism and contrast in the petals and to avoid a “yellow monochrome” tint.
I also made a warm/cold contrast with the colouriser module.
Although it is out of the scope of this post, I also propose a Black and White version.
Greetings from Brussels,
Christian
Yeah I agree, but with filmic v6 this image is busted. Lowering the “brilliance” of the highlights is needed to just snuff out the color break-up and allow the image brightness to be increased a little bit so at least the historgram can go beyond ~60% before the colors start breaking up. So it only looks “mildly too dark”. “increasing brightness at a preserved chrominance” seems to be what is causing heartache for filmic v6 so reducing the brilliance of the highlights backs off the onset of the issue. However the trade-off is a lowered cap on contrast. I don’t know which look I like better, but neither compared to what I see using V5, even if chrominance is not preserved.
Something I don’t understand…
These sunflower pics are quite saturated. If I open one of the jpegs in DT, in this case the filmic-V6-2-75 above, it’s still out of sRGB gamut. How can an sRGB file be out of sRGB gamut? And it’s not as if the pixels are being interpreted as OOG simply because they’re at the maximum value because there are still OOG areas using AdobeRGB as the softproof, and still areas using rec2020 and prophoto. What’s the explanation?
So a few things on this. Make sure when you open a DT-created jpg in DT that you have filmic disabled unless you really want that. Otherwise you get effectively a double dose of tone mapping, with the second one being in display referred space.
The low end out-of-gamut warning is a valid warning in general if you care about it, but it is not indicating it is out of gamut here with a jpg as it should be impossible.
The default warning for low-end out-of-gamut flagging in DT is -4 EV by default, which in terms of powers of 2, 1/2^4 = 1/16 = 0.0625. In terms of, say, HSV, if the image were monochrome, then I think you’d see the warning start at “V” values of 6.25.
If I poke around with the color picker in HSV mode, after converting the jpg image to monochrome, the transitions to flagged areas seem pretty close to “6.25” for V. You can adjust the threshold as low as you want in the warning settings, but if there are “V=0” values somewhere in your jpg, you will never be able to make the warnings go away no matter what the setting is. It’s kind of like Zeno’s paradox .
I don’t know what is going on in the thumbnail, as it doesn’t look like this for me…
hi Brian. Filmic not involved!
I’d always thought the lower and upper thresholds were for luminance rather than gamut. Looking in the manual, the lower one seems to be just luminance whereas the upper one affects the gamut check. To be clear, I was using the gamut button (! within triangle) rather than the clipping button, which I generally have set to lum. only. With this config, using the gamut button, I see no difference in the OOG area with 100% vs. 90% upper threshold. That seems wrong to me.
OOG as shown by the clipping button does indeed vary with the upper threshold. But I’m not sure it’s telling the same story as the gamut button.
I’m afraid you’ve lost me re. the -4EV, V=6.25 etc and I haven’t time tonight to go further. But thanks for your reply.
Oh, yes sorry, I was talking about clipping really. I should not have used the term gamut. You said gamut and heard clipping :). I never use the gamut checker as it never seems to make much sense to me. I’m not sure it actually works like it is supposed to, based on some past comments from others.
and the -4 EV is what I saw with jpgs without me setting it beforehand. It’s different depending on the workspace profile I think.
If you set your display profile working profile and histogram profile to linear rec2020 then you can see true gamut/clipping …your image won’t look right but now you won’t be impacted by I intermediate clipping imposed by your display profile… at least it’s a method you can use to check if that is an issue
.