Thanks heaps mate for taking the time to listen and respond to my rambling. Here's some more
Quote:
Originally Posted by philiphart
... most of the content of my book was written second half of last year and i hadn't even tried LR Timelapse then so was in no position to write about it! have you seen gunther wegner's book as well? probably still doesn't answer your question.
|
I haven't read Gunther ebook yet. Will do that next. I have been logging the changes in XMP files with windiff between saving them from LRT and LR4 and vice versa. It's a bit of a mess at the moment. Obvious version conflicts, etc... I think LR3/CRW6 was stable but LR4/CRW7 are using new processes and LRT doesn't read all of them (yet?). Well it does read the XMP file but the interface doesn't translate it to a usable information so I think the parameters mapping between old and new (2012) may be a little buggy still. Or I am doing things totally wrong. I haven't got much feedback from their forum yet but for the usual 'it's working for me'. There are heaps of posts about metadata issues between LRT and lightroom though.
Quote:
Originally Posted by philiphart
my processing is *much* simpler than yours.. so either the colour issues are there in your files to begin with or all your complicated processing has introduced it. next time you could shoot raw+jpg and see if the colour flicker is there in a jpg sequence. or just export a simple sequence from the RAW files with no tampering of colour at all and see how that looks?
|
The colour flickering is not in the unprocessed RAWS so it is definitely introduced by my processing. Modifying the temperature/exposure sliders even in small increments has a huge impact on the white balance in the histogram from frame to frame. Too much leverage for lack of a better word.
Tone Curve mapping would be more precise on selected keyframes but they just don't pass on to LRT. I can't get them to work anyway. I can see the XMP files loaded back in LRT. Each tone curve (R,G & B) creates an array that is populated with pairs of coordinates (X,Y) that represent each control point on the curve From 0 to 255 on both axis, so minimum of two points (black point and white point) with anything in between. These appear as extra lines in the XMP files within each channel. XMP files are similar to a XML format. But LRT interface columns still reflect what I assume is the old parametric system (as per CRW5 and CRW6): Highlights, Lights, Darks & Shadows sliders in tone curve. No channel separation.
The only few times I shot nightsky stills in JPEG were shocking. Black clipping and under exposed shots. Data that I couldn't recover so I gave up on that real fast. Will try to not color correct a short sequence and see what happens. My last resort would be to batch all the TIFF in PS and run action to get rid of the noise and green cast but it's another step and a lot of big files to handle. Ultimately it might be the solution to not let LRT interpolate colors, just exposure and do the histogram in PS as a final batch. My processing flow needs to be reversed there.
Quote:
Originally Posted by philiphart
daylight white balance should be fine.. that's what i generally use for the dark part of the sequence. i don't do any tweaking over time of white balance/tint/colour tone curves outside twilight. just one setting kept constant throughout dark part of night.
|
Yes daylight white balance was recommended for night as well. I couldn't keep it constant through the night sequence though as I was trying to get rid of that green cast.
Quote:
Originally Posted by philiphart
i dare say the sequence is a little underexposed which probably exacerbates the problem.. lots of colour in the shadow noise. try 30 second frames.. (or a faster lens :-).
|
Yes it is. Trying to hide all the ugly bits.
The lens is 25-18mm. I'm looking for a 12mm or 16mm. I think Alex has a 12mm pancake lens on his. It would certainly make the field more interesting as it gets a bit monotonous in parts.
Quote:
Originally Posted by philiphart
could try desaturating a little, apply colour noise reduction.. hmm.. i'm just about out of ideas. i think the small sensor and pixels in your nex5 are probably your biggest limitation at this point. i don't think your processing skills need any more help!
|
I did push the saturation a little yeah. When I watch the 1080 on my HD TV I went wow! Rio carnival! Too much colors
Maybe I'll calibrate my new PC monitor. I haven't done that yet. Noise is another thing that I haven't approached yet. I'll do it when I have the colours sorted out. Baby steps.
I do need help. If I don't make notes of what I'm doing I forget.
Quote:
Originally Posted by philiphart
[edit.. nex5 pixels look similar size to other cropped sensor dslrs.. but still looks a bit like a sensor performance issue to me..]
cheers
phil
|
Could be too. After all it's not a high end. Just having fun with it.
Alex got his camera spectrum modified so I thik it's more sensitive to red. If I can get these greens down post process and stop the color shifts I'll be a happy camper.
i also like the sunrise sequence over the grass and tents.. very nice light and colour. and the low angle (no tripod!?) actually looks really nice.
I really enjoyed the views, thanks.
