Originally Posted by Peter J. De Smidt
Thank you very much, Peter and Bob; you've given me a lot to chew on for the next few days.
Judging by what you say, I'll have to modify my development.
I just found my old film boxes, and some of the film might have been outdated or improperly stored.
Still, this should give a magenta cast, shouldn't it?
I don't really know how to modify film development to correct a green cast, but I'll have to research that.
Thanks again.
Ari,
I'm not sure what cast your film really has, since, for example, whether it is magenta or green can depend a lot on the settings you use when you scan. The cast could certainly be from older film, or it could be a processing issue. If you look at your blue channel image that you posted, you can clearly see the dark area that starts in the sky and move down into the building. While this could be a film storage issue, it's more likely a processing issue. There are more blotches like that thorough out the image, although they are smaller. They are a bigger defect, imo, than color casts.
To check your developing, get some new film. Take two shots of a scene. Send one to Bob to develop, and then develop the other one yourself. Compare.
Another thing that you can do is to photograph a Macbeth color checker in representative light. Since most color negative film is meant for noon daylight on a clear day, photograph the chart in that type of light. Now develop it. It should be pretty easy now to find out what casts, if any, your materials and processes are giving you. Ideally, the black square on the chart should be in shadow. That would allow you to use the white, gray, and black eye droppers on a curves adjustment layer to neutralize the gray-scale of the image. Usually when the gray-scale of the image is good, the rest of the image will be as well. To shade the black square, I make a rectangular shade using black velvet and a rigid card stock. It goes just around the black square of the chart to keep direct sun off of it.
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
+1. These can be a bit hard to follow, as can his books, but they are worth it.
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
The first problem is that your image doesn't have an embedded profile. For the web, it should be sRGB. When I assign sRGB to your image, I didn't notice a change, but there very well could have been one.
Bright sky: L96, -2a, -6b i.e. probably a slight cyan cast.
Shadow bushes: L12, -3a, 0. i.e. probably a little too blue.
White of tennis shoe: L74, 1a, -4b. i.e. probably a little too blue.
Grass: L63, -9A, 32B, i.e. probably too little cyan or too much yellow.
Ventilation hole in sneakers: L8, 2a, -5B i.e. probably too blue.
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
My numbers agree with Peter.
This image is what I would call close enough for govt work.
The key when neutralizing is to find a good area that is suppose to be white, a mid tone around the L - 50 range that should be neutral and a shadow black area.
When the A & B is 0 or within 1pt that is what I go for in each key area.
Now remember this is only neutralizing and the image may need a visual cast to emphasis the scene and this is where your asthetics take over and make a decision whether the image is dead nuts nuetral or not.
One of the biggest advantages of neutralizing first and formost before any move is to avoid colour curves fringing and crossing with curve shaping and sharpening.
This is why most good scans come in un sharpened and the image is not sharpened until most of your editing moves are completed.
I will make minor bling changes after sharpening but try to keep it to the last step.
As Ari's image shows his input scan is really biased towards green, it should be much closer before PS editing is applied as well not sharpened and at all times try to keep in 16 bit which gives you more headroom for editing.
LAB is not for all workers and there are those here on this Forum that say its bad, I cannot remember a significant file that I have not gone into LAB for some tweaking , curve shaping or sharpening, buts thats my choice and I find it extremely helpful as well very intuitive to colour correction as well as the 0-100 platform for L density.
After a few thousand files I feel very confident with a simple workflow.
colour neutralization in RGB mode
Image weight with Channel Blending in Apply Image -RGB mode
Mode change to LAB - Shadow Highlight if needed on the L Channel
Twisting the L channel to affect Contrast with curves
Twisting the AB channel to colour correct or adjust
Dodging and Burning with Blending modes or curves on the L Channel
Image Sharpening with the L channel using a brush to paint in different sharpening if needed
Mode Change to RGB - final colour blings if needed
Convert to printing profile and print.
Dan M covers all these topics and much much more but the above is what I find appropriate and an image usually takes 5-15 minutes to get what I want.
For most color images, I use a variant of Dan's three step method. 1. get rid of casts. 2. enhance contrast. 3. enhance colors.
In step one, I mainly use an adjustment curve set to "color" blending mode. Use "shadow/highlight" if needed.
Step two involves duplicating the background layer, setting the mode of the new layer to "luminosity" and examining the channels. I replace any bad ones with a better one, for example, I regularly replace the blue channel with the green one using the "apply image" command, and then I use a curve to increase contrast, usually in the green channel, in the most important areas. I regularly use a Hiraloam (high radius low amount) sharpening on one of the channels.
Step three involves changing the image mode to LAB and making a new layer set to overlay or multiple and adding a luminosity mask to the layer.
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
Thanks for that Bob. I can only find it in italian. I haven't got photoshop so I don't want to subscribe just yet to kelby's site but it looks good so I've booked marked it just in case Santa brings me Photoshop later in the year. I don't suppose there's a free english language version somewhere?
Thanks Peter. I think I've found something that allows me to monitor certain points of an image as I alter the RGB channels. I'd be grateful if you'd take a look at the image attached to see if I'm on the right path. I've adjusted colour at 3 points:
1) the brightest part of the sky
2)the grey cuff of the shoe
3) the shoe's ventilation hole.
Do they seem to have been improved compared to the first image? I haven't tried to alter colour elsewhere in the image.
I don't wish to be a pain.
Oh, and is there an embedded profile now?
Pete
Attachment 62174
No, there doesn't seem to be an embedded profile. The image does look better, though.
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
I think you're dealing with a combination of a processing problem and a scanning problem.
As others have pointed out, there are some blotches or stains on the negative. The doesn't necessarily indicate that the processing is causing a color cast, but it's certainly not making your job any easier.
The scan looks clipped in the shadows to me. I didn't put it in PS and I'm on my laptop--but can you see the entire histogram in each color channel, or are they clipped? The shadows look blocked up and if I had to guess I'd say that the green channel may be clipped on the highlights as well. (To check for sure remove the white border.)
Again, you can often correct a scan with clipped shadows or highlights, but it can be more difficult.
I've never used epson scan. But in general I scan the way you mentioned, I look at the histograms and I pull in the points to where the curve starts for the highlights and shadows for each channel. One thing I've noticed is that if you're scanning the film edge or if there are clip marks on your film, it can throw off the histogram. So when I make my settings I create a crop selection box that is all image area and excludes the film edge, clip marks, etc. Once I set my points I can reset the crop to scan the entire film.
That usually gets me pretty close. I sometimes use the LAB curves in my scan software (Trident), but mostly just the L channel to make sure I have a nice distribution of tones for printing.
It may be that the Epson software is doing some additional processing or it may be that it's just not accurate. If you scans are getting clipped, you may need to not pull the points in so far. The key is to get a good scan that contains as much information as possible. My scans are usually a little flat, but not too much. I try to spread out the histogram as much as possible without clipping in any of the color channels.
In PS I color correct by again setting the white and black points for each channel, and I sometimes add a minor curves adjustment to boost contrast. That's usually about it.
I'd try scanning a negative processed at a good lab to make sure you scanning routine is working, then you can tackle the processing issues.
Well, it's perhaps a slight improvement, certainly in terms of time, but it's still not perfect.I've posted it here
Any input welcome.
Pete
Posting Permissions
Reply With Quote
Bookmarks