-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
How To Get Good Images with D7
- Thread starter Jawed49152
- Start date
Jawed49152
Senior Member
Ok, I have posted some small samples on my webby:
DIVU images processed normally:
http://www.cupidity.force9.co.uk/Photos/D7Demo/DIVUOriginal/FrameSet.htm
DIVU images using my correction curve:
http://www.cupidity.force9.co.uk/Photos/D7Demo/DIVUCorrection/FrameSet.htm
Images loaded in PS assuming my monitor's profile:
http://www.cupidity.force9.co.uk/Photos/D7Demo/Monitor/FrameSet.htm
All images are full-frame.
I recommend, if you have the screen space, that you view one image in three versions side by side... Ideally they should be loaded into Photoshop. They are tagged with the sRGB profile, so they will display vaguely well in your browser, but Windows 95 and 98 have problems with this (e.g. yellows are too green). So use PS if you can. Or Paint Shop Pro set to use the sRGB space.
Curiously enough, making these images this small reduces the contrast differences between them and seems to selectively de-saturate some colours - weird. I didn't really discover this until I'd finished making the files (a fair amount of work!). Normally I post web samples at much larger image sizes, but because I've posted so many images, and because some peeps will prolly have small screens, I've kept the images small.
There are a few of the monitor profile images where I've used too much saturation, to correct for using the monitor profile. e.g. 2189 and 2271. Darn.
Finally it's worth pointing out that the "monitor" images are roughly 50-90% larger than their DIVU counterparts (all images saved with quality level 9 from Photoshop, which is high quality). This would seem to indicate they have the most detail!
See if you can spot the Phil-inspired images!
Jawed
DIVU images processed normally:
http://www.cupidity.force9.co.uk/Photos/D7Demo/DIVUOriginal/FrameSet.htm
DIVU images using my correction curve:
http://www.cupidity.force9.co.uk/Photos/D7Demo/DIVUCorrection/FrameSet.htm
Images loaded in PS assuming my monitor's profile:
http://www.cupidity.force9.co.uk/Photos/D7Demo/Monitor/FrameSet.htm
All images are full-frame.
I recommend, if you have the screen space, that you view one image in three versions side by side... Ideally they should be loaded into Photoshop. They are tagged with the sRGB profile, so they will display vaguely well in your browser, but Windows 95 and 98 have problems with this (e.g. yellows are too green). So use PS if you can. Or Paint Shop Pro set to use the sRGB space.
Curiously enough, making these images this small reduces the contrast differences between them and seems to selectively de-saturate some colours - weird. I didn't really discover this until I'd finished making the files (a fair amount of work!). Normally I post web samples at much larger image sizes, but because I've posted so many images, and because some peeps will prolly have small screens, I've kept the images small.
There are a few of the monitor profile images where I've used too much saturation, to correct for using the monitor profile. e.g. 2189 and 2271. Darn.
Finally it's worth pointing out that the "monitor" images are roughly 50-90% larger than their DIVU counterparts (all images saved with quality level 9 from Photoshop, which is high quality). This would seem to indicate they have the most detail!
See if you can spot the Phil-inspired images!
Jawed
Stephen Lau
Forum Enthusiast
Jawed,
I like the PS/Monitor profile version the best. It is relatively the most contrast of the three versions. But for this diffused lighting, it works the best, at least on my monitor. The Normal DIVU is the second best and the DIVU/Correction curve image seems to washed out and too low in contrast. I do not notice any color difference from the three versions though.
Same old story again. With different lighting and contrast of a scene, my preference may not be the same (most likely). Now, you have three tools and use them accordingly.
I like the PS/Monitor profile version the best. It is relatively the most contrast of the three versions. But for this diffused lighting, it works the best, at least on my monitor. The Normal DIVU is the second best and the DIVU/Correction curve image seems to washed out and too low in contrast. I do not notice any color difference from the three versions though.
Same old story again. With different lighting and contrast of a scene, my preference may not be the same (most likely). Now, you have three tools and use them accordingly.
Nicole62214
Leading Member
Same for me most of the time, but divu is also well
I will compare the divu and the one with the monitor profile
0029 the mauve are more saturated in divu skin is better in monitor profile
0166 more detail (brick of the house) in the monitor profile
0434 monitor look more sunny (the plant) stores are more visible in divu
0439 divu more saturated
0443 more details in the monitor profile
0446 in monitor the red dress is more orange than divu: very different
0455 divu more saturated and brillant
0625 divu house is more tinted
0743 divu a litle darker than the other, so the yellow is stronger
1747 same thing in the red than for the 0446
1995 monitor is really much better particulary for the skin
2131 the divu look a very very little bit better
2135 they are very close together
2189 colors a little stronger in divu
2271 very much better in monitor. Colors are more interesting
2276 like 0029. Mauve more saturated in divu
2289 skin is always better in monitor profile. The jeans are very different
2382 more detail in monitor
2389 the sky in divu is more "menaçant" darks are better in divu but clearer objetcs like the fence is better in monitor
2439 are very closed divu a little bit stronger
You made a colossal work. Thank you for the big time you passed to share these treasures.
My conclusions are that colors are a little bit more saturated in divu, so when you have more contrast exemple for the skin in some of the picts it is a little bit too dark, and this is why they are better in monitor. The only thing to know right now for me is: are the colors too saturated in divu or not enough in monitor profile. Exemple the reds, particulary in the pict 0446 appear to me very different.
Nicole
henri56123
Leading Member
Hi,
I work on a Mac G4 with a Sony monitor; very recently I found out that the color management of Mac's and PC's are so different that results on one system are not always applicable to the other.
I worked out a very good system on my Mac to convert the camera pictures to Photoshop working profile using DIVU as the conversion tool. The results are absolutely great: fine colors, better shadow detail, an overall improvement in perception of sharpness; that works best in daylight scenes, preferably with sunshine. however shots from the same area taken later in the day, just before the sun will go down, start showing (with this conversion) a kind of mauve tint, which is easily removed removed in Photoshop^
The point is : contrarily to scanners, digital camera do not have a single color space, it varies with the light conditions , and there is no single profile that works perfect for all conditions (a profile is a description of the color space) - this statement is confirmed in professional literature; all we can do is find the best working compromise.
Henri
I work on a Mac G4 with a Sony monitor; very recently I found out that the color management of Mac's and PC's are so different that results on one system are not always applicable to the other.
I worked out a very good system on my Mac to convert the camera pictures to Photoshop working profile using DIVU as the conversion tool. The results are absolutely great: fine colors, better shadow detail, an overall improvement in perception of sharpness; that works best in daylight scenes, preferably with sunshine. however shots from the same area taken later in the day, just before the sun will go down, start showing (with this conversion) a kind of mauve tint, which is easily removed removed in Photoshop^
The point is : contrarily to scanners, digital camera do not have a single color space, it varies with the light conditions , and there is no single profile that works perfect for all conditions (a profile is a description of the color space) - this statement is confirmed in professional literature; all we can do is find the best working compromise.
Henri
Thanks for your comments Nicole. You have, like me, identified some occasional advantages for DIVU. I prefer Monitor for the way it treats skin and you seem to agree with me.
As far as colour is concerned:
1. the pillar box red of 1747 is orange-red because evening sunlight is shining very strongly on buildings across the street and reflecting back. You can see a shadow on the ground going "the wrong way". This colour and light was the reason I took the picture. For me Monitor has captured the colour best.
2. In 439 and 2189 the yellows have more green in them than I think they should. This is a problem that viewing on the web makes much worse. If you view these images in Photoshop there is hardly any difference .
3. Some blues are also quite different. I have discovered that blue (sky, usually) can often change to a kind of mauve colour when it is viewed on the web (Internet Explorer in Windows 95 and 98). This is quite frustrating as some of my pix posted on the web suffer from this problem quite a lot. The Monitor profile appears to reduce this problem.
In general, with colour, it's all a little bit "random". The human eye adapts to colour very readily, so for example, we don't see the blueness that the bright midday sky puts into most scenes, until we look at the picture (this is why I leave the camera in Overcast white balance, as blueness from the sky is something I dislike). Also, if the picture is under-exposed then this will tend to increase the saturation.
The DIVU images with my DIVUCorrection curve demonstrate that simply by changing contrast you can make a lot of changes to perceived saturation, to the extent that DIVUCorrection shows colours that are the least like normal DIVU with Monitor being a better match for DIVU (there are some exceptions). If I use LAB mode in Photoshop to change contrast these colour changes are dramatically reduced.
So I use people's skin as the most demanding test. It's rare that people really look like they do in the DIVU versions. Also, in a lot of pictures, DIVU makes trees blur together, so that you can't tell where one tree ends and another one starts (see 2276 and 2382) which is weird.
Occasionally, an image produced by the Monitor process seems a little too dark in the darkest shadows, hiding detail that DIVU shows. This can be fixed. Indeed DIVU images can be fixed to remove their muddiness. The DIVUCorrection curve is a little too bright (I said this when I first published it), but it can also be tweaked, to taste.
So I have a set of choices as to where I start. Because I am fairly expert with Photoshop curves, I can get what I want with any of these versions. What's nice is that the Monitor version seems to come closest to what I want from my pictures and I'm naturally quite a lazy person...
Jawed
As far as colour is concerned:
1. the pillar box red of 1747 is orange-red because evening sunlight is shining very strongly on buildings across the street and reflecting back. You can see a shadow on the ground going "the wrong way". This colour and light was the reason I took the picture. For me Monitor has captured the colour best.
2. In 439 and 2189 the yellows have more green in them than I think they should. This is a problem that viewing on the web makes much worse. If you view these images in Photoshop there is hardly any difference .
3. Some blues are also quite different. I have discovered that blue (sky, usually) can often change to a kind of mauve colour when it is viewed on the web (Internet Explorer in Windows 95 and 98). This is quite frustrating as some of my pix posted on the web suffer from this problem quite a lot. The Monitor profile appears to reduce this problem.
In general, with colour, it's all a little bit "random". The human eye adapts to colour very readily, so for example, we don't see the blueness that the bright midday sky puts into most scenes, until we look at the picture (this is why I leave the camera in Overcast white balance, as blueness from the sky is something I dislike). Also, if the picture is under-exposed then this will tend to increase the saturation.
The DIVU images with my DIVUCorrection curve demonstrate that simply by changing contrast you can make a lot of changes to perceived saturation, to the extent that DIVUCorrection shows colours that are the least like normal DIVU with Monitor being a better match for DIVU (there are some exceptions). If I use LAB mode in Photoshop to change contrast these colour changes are dramatically reduced.
So I use people's skin as the most demanding test. It's rare that people really look like they do in the DIVU versions. Also, in a lot of pictures, DIVU makes trees blur together, so that you can't tell where one tree ends and another one starts (see 2276 and 2382) which is weird.
Occasionally, an image produced by the Monitor process seems a little too dark in the darkest shadows, hiding detail that DIVU shows. This can be fixed. Indeed DIVU images can be fixed to remove their muddiness. The DIVUCorrection curve is a little too bright (I said this when I first published it), but it can also be tweaked, to taste.
So I have a set of choices as to where I start. Because I am fairly expert with Photoshop curves, I can get what I want with any of these versions. What's nice is that the Monitor version seems to come closest to what I want from my pictures and I'm naturally quite a lazy person...
Jawed
henri56123
Leading Member
From what I see, you don't seem to be the laziest person
By the way: I also do more and more editing in the LAB mode: color artifarcts, noise, sharpening all in one go; much better than before
regards
henri
Stephen Lau
Forum Enthusiast
Henri,
Would you please get into more details of editing in LAB mode? What is the advantage and disadvantage of using LAB mode? My undertstanding (which is not much) is the LAB mode converts and restricts the images color space. I have no experience in the LAB mode, except occasionally use it to minimize noise. TIA.
Would you please get into more details of editing in LAB mode? What is the advantage and disadvantage of using LAB mode? My undertstanding (which is not much) is the LAB mode converts and restricts the images color space. I have no experience in the LAB mode, except occasionally use it to minimize noise. TIA.
My present procedure is to convert images to raw mode in PS6 before doing anything to them. I then apply any curves, hue sat presets, etc, and adjust the final contrast of the image. Then I convert back to RGB for printing. I find that LAB gives me control over luminosity with minimal or no color shifting. Having said that, there are many images that require so little change that I sometimes do everything in RGB.
dh
dh
Stephen Lau
Forum Enthusiast
dh,
I am still searching for a workflow for myself. After all these profiling, and went around in circles a few times, I knid of settle down a bit. Mostly I just open the jpeg file in PS6 and Leave As Is (no color match) and it seems to work good for me for most images. Assigning Raw profile is my last resort when I have an image too contrast. Don't you find assigning the Raw profile creates way too flat an image in most cases? I tried that and I have to make lots of correction (color and contrast) and the final image seems to suffer throguh the process. Is LAB the one precess I am missing?
I have to investigate more in LAB and pay more attention to to the color shift issue. There are so many ways to skin a cat!!! Who was the one said that digital camera can provide faster feedback :> )) ???
I am still searching for a workflow for myself. After all these profiling, and went around in circles a few times, I knid of settle down a bit. Mostly I just open the jpeg file in PS6 and Leave As Is (no color match) and it seems to work good for me for most images. Assigning Raw profile is my last resort when I have an image too contrast. Don't you find assigning the Raw profile creates way too flat an image in most cases? I tried that and I have to make lots of correction (color and contrast) and the final image seems to suffer throguh the process. Is LAB the one precess I am missing?
I have to investigate more in LAB and pay more attention to to the color shift issue. There are so many ways to skin a cat!!! Who was the one said that digital camera can provide faster feedback :> )) ???
henri56123
Leading Member
Hi Stephen
Basically the LAB mode is a better represenation of how our brains process the information from our eyes by separation luminance and color information whereby luminance is the most important info, color a detail ; (this should be the main reason why we like B&W pictures) ther is a good article - see
http://www.luminous-landscape.com under Light, color and human vision
But in practical terms: LAB separates luminance from color; that allows you to eliminate color artifarcts in the color channels without affecting the luminance info and to do things like noise elimination and above all sharpening without affecting the color information
Here is a procedure that works very well (much better than simple unsharp mask)
convert the pic to Lab mode
select channel a and apply gaussian blur (3-6 pixels)
select channel b and apply gaussian blur (a little more than in a)
this takes care of color artifarcts
copy the lightness channel and select it
Filter - Stylise - Find edges
apply Gaussian blur 2-4pixels
this will serve as a selection
re-select the Lightness channel
select - Load selection - Lightness copy
filter Median 1 to 2 pixels
this takes care of noise
inverse the Lightness copy selection
this is the most important part and you may want to make the selection invisible (View - Show - Slection edges)
filter unsharp mask - here you can do extremes like radius 1 pixels and 500 % or radius 10-20 pixels and 2-400 %, it really depends on the picture and you have to look carefully; note that you only affect the "edges" but you need to look well; never use threshold
return to rgb
Henri
Basically the LAB mode is a better represenation of how our brains process the information from our eyes by separation luminance and color information whereby luminance is the most important info, color a detail ; (this should be the main reason why we like B&W pictures) ther is a good article - see
http://www.luminous-landscape.com under Light, color and human vision
But in practical terms: LAB separates luminance from color; that allows you to eliminate color artifarcts in the color channels without affecting the luminance info and to do things like noise elimination and above all sharpening without affecting the color information
Here is a procedure that works very well (much better than simple unsharp mask)
convert the pic to Lab mode
select channel a and apply gaussian blur (3-6 pixels)
select channel b and apply gaussian blur (a little more than in a)
this takes care of color artifarcts
copy the lightness channel and select it
Filter - Stylise - Find edges
apply Gaussian blur 2-4pixels
this will serve as a selection
re-select the Lightness channel
select - Load selection - Lightness copy
filter Median 1 to 2 pixels
this takes care of noise
inverse the Lightness copy selection
this is the most important part and you may want to make the selection invisible (View - Show - Slection edges)
filter unsharp mask - here you can do extremes like radius 1 pixels and 500 % or radius 10-20 pixels and 2-400 %, it really depends on the picture and you have to look carefully; note that you only affect the "edges" but you need to look well; never use threshold
return to rgb
Henri
I just did a few experiments with a shot of 30 colorful magic markers covering the spectrum at various color level settings on the camera. It looks like shooting at -2 or -3, then processing thru Divu with colormanagement on gets you almost identical hue and saturation levels as opening untagged files in Photoshop shot at color level 0.
I also tried conversions to my SRGB Photoshop workspace to see if opening untagged data resulted in missing color, and also to see if any process could scale camera color info into a smaller color space better. I thought BruceRGB as the assumed working space worked better than AdobeRGB since the latter hyped up neutral colors unrealistically if the exposure had a lot of contrast. Both worked better than directly converting to SRGB. BruceRGB also agreed most with the Divu processed file. I also tried both Minolta profiles, and didn't get any color that reminded me more of the markers in front of me.
No matter what process I used to get to SRGB, the pictures tended to miss on extremely bright magentas and oranges and to a lesser extent on greens blues and violets, and also the errors were almost identical. I set up the markers right next to my monitor so I could compare with the original. I'm fortunate enough to have a colormeter to calibrate my monitor, so I think I'm pretty accurate. It looks to me as if the camera might use BruceRGB as the files' target color space as some scanners do since it's supposed to be a good space for converting to print.
Don't know if it's true, but maybe it gives people some more clues. The fun implication is that if you don't want to spend money on expensive graphics software and calibration tools for your monitor, you could get one of those relatively inexpensive inkjetmall custom profiles for your printer and be doing pretty well as long as you have the courage not to look at your monitor and trust Divu.
I also tried conversions to my SRGB Photoshop workspace to see if opening untagged data resulted in missing color, and also to see if any process could scale camera color info into a smaller color space better. I thought BruceRGB as the assumed working space worked better than AdobeRGB since the latter hyped up neutral colors unrealistically if the exposure had a lot of contrast. Both worked better than directly converting to SRGB. BruceRGB also agreed most with the Divu processed file. I also tried both Minolta profiles, and didn't get any color that reminded me more of the markers in front of me.
No matter what process I used to get to SRGB, the pictures tended to miss on extremely bright magentas and oranges and to a lesser extent on greens blues and violets, and also the errors were almost identical. I set up the markers right next to my monitor so I could compare with the original. I'm fortunate enough to have a colormeter to calibrate my monitor, so I think I'm pretty accurate. It looks to me as if the camera might use BruceRGB as the files' target color space as some scanners do since it's supposed to be a good space for converting to print.
Don't know if it's true, but maybe it gives people some more clues. The fun implication is that if you don't want to spend money on expensive graphics software and calibration tools for your monitor, you could get one of those relatively inexpensive inkjetmall custom profiles for your printer and be doing pretty well as long as you have the courage not to look at your monitor and trust Divu.
henri56123
Leading Member
Lee Hurd wrote:
Hi, I have 2 questions
henri
Hi, I have 2 questions
- what is BruceRGB ?
henri
Henri,
I did a Google search and found this:
This month BruceRGB celebrates its second birthday. For those who haven't heard of it, BruceRGB is my attempt to provide the world with a reasonably safe RGB working space for use with Adobe Photoshop 5.0 and later.
On: http://www.creativepro.com/author/home/40.html
Bruce is Bruce Fraser a photoshop book author.
Marty
I did a Google search and found this:
This month BruceRGB celebrates its second birthday. For those who haven't heard of it, BruceRGB is my attempt to provide the world with a reasonably safe RGB working space for use with Adobe Photoshop 5.0 and later.
On: http://www.creativepro.com/author/home/40.html
Bruce is Bruce Fraser a photoshop book author.
Marty
MartyB got it. I remember when Microsoft decided everyone should use SRGB as the color standard, and the Mac guys wanted the Mac standard because it was better for print, but SRGB won out. I think Fraser proposed his color space as an improvement over the shortcomings of SRGB which showed up in my Magic Marker test. I forgot to mention I switched my working space to BruceRGB and saw all the colors reproduced very well.
I missed the second question. SRGB is usually used for games, multimedia and web destined stuff since everyone who has windows is supposed to be using it. That's where a lot of my work goes so I use it most of the time.
henri56123
Leading Member
Thank you forlks, I'ml gonna read this Bruce F site tonight, looks very interesting - one can always learn something
Henri
Henri
Russell Bozian
Well-known member
Thanks Marty, a huge amount of good material here for those of us trying to understand colorspaces. And quite relevant to anyone doing image editing, whether they use Photoshop or not, even though this guy is a big Photoshop dude.
Keyboard shortcuts
- f
- Forum