Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: Loss of detail in tulip photos - advice?  (Read 3704 times)
kdern
Newbie
*
Offline Offline

Posts: 19


« on: April 21, 2013, 10:02:53 PM »
ReplyReply

Can anyone help me with this? I am currently using a Fuji x100s but I have had this issue with other cameras as well.

When I shoot very colorful flowers, like tulips, they just look like blobs of color with very little detail. I just shot some colorful tulips today, on a bright but cloudy day (no direct sun), in both JPG and RAW, and although the RAW has more detail than the JPG, neither has a ton of detail.

Check out these comparison - JPG on the left, RAW on the right, screenshots from Lightroom. The green leaves of the flowers look great, but the flowers themselves look like I have turned the saturation way up.

These are straight out of the camera - no processing at all.

Screenshot, zoomed to 1:2 Larger image here.


Another, zoomed to 1:4 Larger image here.


One more, zoomed to 1:2 Larger image here


Any suggestions? I assume it's something to do with my exposure settings?

I wonder if changing Dynamic Range settings would make a difference? I had mine at DR100.

Note: the screenshots are showing some distortion on the flowers. I'm not seeing that in Lightroom. Let me know if you need me to post some higher res images online.

Thank you!
Logged
Glenn NK
Sr. Member
****
Offline Offline

Posts: 281


« Reply #1 on: April 21, 2013, 11:38:08 PM »
ReplyReply

Don't feel as though you are the only one with this problem.  I have 20,000 to  30,000 flower images, and getting detail can be difficult.

Assuming that you are using Lightoom - don't know what options you have with a Fuji camera - but I usually use either Camera Faithful or Camera Neutral for the Profile.   Landscape usually blows the details right out of bright colours (I'm using Canon bodies).

Also be careful with Contrast - it can be problematic.

I'm not convinced that in camera JPEGs are faithful to all colours and I want total control over Hue and Saturation, so I only use RAW.

It's very easy to blow out the reds - the RGB histogram will help  (the Brightness histogram is pretty well useless).  Actually it's easy to blow out any of the colours in flowers (I'm very careful with purple flowers - learned the hard way).

Glenn
« Last Edit: April 21, 2013, 11:48:34 PM by Glenn NK » Logged

Economics:  the study of achieving infinite growth with finite resources
k bennett
Sr. Member
****
Offline Offline

Posts: 1434


WWW
« Reply #2 on: April 22, 2013, 06:35:16 AM »
ReplyReply

You can try painting in some Clarity with a brush around the tulip petals. This might help define the edges.

Also realize that flowers have all kinds of UV-related issues, and while your overall exposure might be perfect (it looks fine in your samples) the blossoms can still have some color channels blown out.
Logged

Equipment: a camera and some lenses.
John Cothron
Full Member
***
Offline Offline

Posts: 170



WWW
« Reply #3 on: April 22, 2013, 07:27:47 AM »
ReplyReply

Use the HSL sliders and see if you can pull down the Sat/Lum on those flowers to bring back the detail enough to get some detail but not lose the color.  It's worked for me a few times.  Or.. bring down the whole scene till the flowers look more to your taste, then take everything else up using the sliders.  First method would probably be easiest.
Logged

stamper
Sr. Member
****
Online Online

Posts: 2651


« Reply #4 on: April 22, 2013, 07:40:29 AM »
ReplyReply

Try spot metering in future. In evaluative/matrix metering the flowers will usually be brighter than the background which means they are in danger of becoming overexposed. If you use PS then Dan Magulis has a answer for the problem though it is fiddly to achieve. In PS look at all of the separate channels, pick out the one with the most detail ( usually the green channel ) and channel blend it into the composite file using luminousity.

http://www.fredmiranda.com/forum/z/665853/0
Logged

bjanes
Sr. Member
****
Offline Offline

Posts: 2783



« Reply #5 on: April 22, 2013, 08:06:04 AM »
ReplyReply

Can anyone help me with this? I am currently using a Fuji x100s but I have had this issue with other cameras as well.

When I shoot very colorful flowers, like tulips, they just look like blobs of color with very little detail. I just shot some colorful tulips today, on a bright but cloudy day (no direct sun), in both JPG and RAW, and although the RAW has more detail than the JPG, neither has a ton of detail.

Check out these comparison - JPG on the left, RAW on the right, screenshots from Lightroom. The green leaves of the flowers look great, but the flowers themselves look like I have turned the saturation way up.

Any suggestions? I assume it's something to do with my exposure settings?

I wonder if changing Dynamic Range settings would make a difference? I had mine at DR100.

Note: the screenshots are showing some distortion on the flowers. I'm not seeing that in Lightroom. Let me know if you need me to post some higher res images online.

Thank you!
Loss of detail in the saturated colors can result from saturation or luminance clipping. To help prevent the former, the use of ProPhotoRGB is helpful. When working from raw (in camera JPEGs AFAIK do not support ProPhotoRGB), saturation clipping in ACR and LR is indicated by colored spikes at the right of the histogram. If these appear, render into ProPhotoRGB. If that does not eliminate the problem, further adjustments may be needed as suggested below and by others in this thread. Softproofing is a definite help here, especially if you have a wide gamut monitor.

Of course, it is essential not to overexpose. Some degree of underexposure can help, since RGB color spaces can encode less saturation at high luminances (the plot is conical as shown below. Luminance is represented by the Y axis and saturation by the X-axis).

The final consideration is the output device. Wide gamut monitors and printers help. If saturation clipping occurs, you may have to reduce saturation and/or luminance.

Regards,

Bill
« Last Edit: April 22, 2013, 08:14:00 AM by bjanes » Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2108


« Reply #6 on: April 22, 2013, 11:35:55 AM »
ReplyReply

The flowers are probably clipping in your monitor display space, so they are getting mapped to a solid color.  You can verify this by using soft proofing (assuming you're using v4 of Lr or higher).  For strongly-colored bulbs like these, you can try reducing one or more of the primary saturation sliders in the Camera Calibration panel (e.g., red, green, and/or blue saturation).
Logged

richarddd
Jr. Member
**
Offline Offline

Posts: 86


« Reply #7 on: April 22, 2013, 12:01:27 PM »
ReplyReply

The flowers are probably clipping in your monitor display space, so they are getting mapped to a solid color.  You can verify this by using soft proofing (assuming you're using v4 of Lr or higher).  For strongly-colored bulbs like these, you can try reducing one or more of the primary saturation sliders in the Camera Calibration panel (e.g., red, green, and/or blue saturation).
What's the difference between doing that and reducing saturation in one or more colors in the HSL panel?
Logged

jrp
Jr. Member
**
Offline Offline

Posts: 77


« Reply #8 on: April 22, 2013, 01:04:44 PM »
ReplyReply

Take a muted copy of the pic into Photoshop, copy the background layer,  blend a % of one of the other channels into the red channel in darken mode.   You can now increase the contrast of the brighter end of the new red channel and change the new layer to luminosity mode.  This can give you a bit more texture in the red channel, and you can boost the reds back.

Logged
fike
Sr. Member
****
Offline Offline

Posts: 1373


Hiker Photographer


WWW
« Reply #9 on: April 22, 2013, 01:14:17 PM »
ReplyReply

Take a muted copy of the pic into Photoshop, copy the background layer,  blend a % of one of the other channels into the red channel in darken mode.   You can now increase the contrast of the brighter end of the new red channel and change the new layer to luminosity mode.  This can give you a bit more texture in the red channel, and you can boost the reds back.



that is a really great idea for an everpresent problem.  I find that extreme yellows, oranges, and reds (almost always found in flowers) are where I run into this the most. I frequently find myself being much more conservative about exposure and ETTR on flower pictures. It is important to set your histogram so that you can see each of the three different channels (usually called something like RGB histogram). Sometimes you will have bad clipping in one critical color that doesn't seem so problematic when viewed on the luminosity histogram.
Logged

Fike, Trailpixie, or Marc Shaffer
marcshaffer.net
TrailPixie.net

I carry an M43 ILC, a couple of good lenses, and a tripod.
madmanchan
Sr. Member
****
Offline Offline

Posts: 2108


« Reply #10 on: April 23, 2013, 09:05:56 AM »
ReplyReply

What's the difference between doing that and reducing saturation in one or more colors in the HSL panel?

The difference is the image processing stage at which the clipping is taking place.  If the clipping is happening at the camera color profile stage, that needs to be addressed first, because if colors clip at that stage nothing that you do with HSL or Saturation or Vibrance will help bring back the color detail.  If the clipping is happening instead at the monitor display profile stage, then you can use any of these color controls to help bring back color detail.
Logged

langier
Sr. Member
****
Offline Offline

Posts: 631



WWW
« Reply #11 on: April 23, 2013, 09:58:26 AM »
ReplyReply

When you shoot, you need to watch the Red histogram. The flowers are oversaturatiing the red channel, thus the lack of detail. You may need to under expose when you are shooting. Try bracketing at minus 1/2, 1, 1.5, 2 stops next time and see how each of the three histograms loom.

For your existing images with ACR or LR, go to the individual color sliders and reduce the luminance (brightness) and probably saturation of the Red and perhaps Orange and Magenta "channels." You may loose a little of the "pop" of the colors but there maybe the detail you are missing.

I have this same issue when shooting sunflowers and many paintings that have saturated reds. I tend to look at the global histogram and not at the RGB set. By pulling back on the sliders a little in post on the luminance and sometimes on the saturation this takes care of the situation pretty well.

You might also shoot the Color Checker and run a custom profile for your camera and lighting. It gets you closer to overall balance at the expense of personal and individual color palate but for some images may make processing easier.
Logged

Larry Angier
ASMP, NAPP, ACT, and many more!

Webmaster, RANGE magazine
Editor emeritus, NorCal Quarterly

web--http://www.angier-fox.photoshelter.com
facebook--larry.angier
twitter--#larryangier
google+LarryAngier
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1153



WWW
« Reply #12 on: April 24, 2013, 07:25:09 AM »
ReplyReply

Here's a demonstration of the type of edits I usually have to apply to similarly clipped flower shots shooting Raw.

Underexposing the shot didn't help. Applying a custom DNG camera profile made it worse so I had to resort to default ACR 4.4 profile and applied similar edits shown below. I have some flower shots lit by direct sunlight that are far worse than yours and I recovered them to have them look better than what's shown in the edited results on your uploaded jpeg shown below.

As to the cause of why this happens I can only guess from what I've learned about digital sensors that it may have something to do the quality/thickness and type of sensor filters that control/affect certain spectrums of light that causes a non-linear response.
« Last Edit: April 24, 2013, 07:32:01 AM by tlooknbill » Logged
Rory
Full Member
***
Offline Offline

Posts: 159


« Reply #13 on: April 24, 2013, 10:10:14 AM »
ReplyReply

The difference is the image processing stage at which the clipping is taking place.  If the clipping is happening at the camera color profile stage, that needs to be addressed first, because if colors clip at that stage nothing that you do with HSL or Saturation or Vibrance will help bring back the color detail.  If the clipping is happening instead at the monitor display profile stage, then you can use any of these color controls to help bring back color detail.

Excellent point Eric.  Thanks very much for your expertise!
Logged
elied
Sr. Member
****
Offline Offline

Posts: 262


« Reply #14 on: April 25, 2013, 05:43:36 AM »
ReplyReply

One of the major culprits in causing red channel clipping when photographing flowers in sunshine has not been mentioned in this thread, and that is WB. Sensors are most sensitive to green. Their response to light decreases as the wavelength decreases, which means they are naturally least sensitive to blue and would normally be most sensitive to red, however that high sensitivity to red means they are also highly affected by IR radiation. An anti-IR filter in front of the sensor also cuts into the exposure to red, which leaves green filtered sensels as the most exposed. Thus WB has a dual purpose; to boost the red and blue channels to parity with the green and then to further boost them to compensate for the color of the light. Daylight is blue so WB for daylight gives the bigger push to the red channel. Typically, it will more than double the red values and reds that were not clipped in the Raw capture will become clipped. The photo below,

has a Raw histogram like this:

The red channel has a maximum value of 7,399 (14 bit) but WB will increase that to above the clipping point (which for the Canon 5D2 is 14,736 after black subtraction). This is what will happen, for instance, with in-camera processing to jpg.

However, because the full, unclipped red data is there in the Raw, intelligent processing with LR will retain it. First of all, for a photo like this where most of the frame is occupied by the flower, the light that is entering the lens is not really natural daylight. It is daylight as reflected from the flower which has absorbed a lot of the blue component and is sending to the camera a much more reddish light. This means that daylight WB, in the range 5,000 to 5,500 is not really appropriate. I set it for around 4,000 and then tweak for the color I want. This has the benefit of reducing the amount of red boost (and increasing blue, but that doesn't bother me). Often this in itself will be enough to prevent clipping. If not, then the next step is to reduce Highlights and Exposure which globally reduces all three channels, of course, and then restoring the green and blue channels to taste.

A second problem, unique to LR but solved in LR4 with the inclusion of soft proofing, is that the LR histogram displays values relative to a very large gamut. Highlight values that seemed safely far from clipping might clip in a conversion to the small gamut of sRGB. For instance a value that might be 240 in a ProPhoto jpg will be 255, i.e. clipped, in sRGB. The use of soft proofing will predict this.
Logged

Roll over Ed Weston,
Tell Ansel Adams the news
bjanes
Sr. Member
****
Offline Offline

Posts: 2783



« Reply #15 on: April 25, 2013, 08:53:39 AM »
ReplyReply

One of the major culprits in causing red channel clipping when photographing flowers in sunshine has not been mentioned in this thread, and that is WB.

Elied,

An excellent post. As you point out, WB clipping can be eliminated by using WB multipliers of less than one (normally the green WB multiplier is 1.0 and the red and blue multipliers are > 1.0). Guillermo Luijk explains this clearly in his treatise on DCRaw (interested readers should refer to the section on WB and highlights in his post).

The problem of clipping with WB can be partially addressed by using the RGB histograms on most newer DSLRs. However, since the widest color space that most cameras offer for their JPEG reviews and histograms is AdobeRGB and saturation clipping can still occur. This is illustrated by a shot of a yellow flower with saturated green and red. Exposure 10 is fully ETTR without clipping as shown by the Rawdigger histogram, but the red and green channels are clipped on the camera histogram. This is due to the conservative nature of the camera highlight warning which allows about 0.5 EV of highlight headroom and by saturatioin clipping in the AdobeRGB space for which the camera was set. Reducing exposure eliminates the green clipping, but the red clipping can't be eliminated even with drastic underexposure due to saturation clipping.



With ACR 7.4 and PV2012 and rendering into ProPhotoRGB with nominal exposure (-0.5 EV to allow for the Exposure Offset that ACR and LR use for the D3), the red is strongly clipped (the red multiplier is 1.70 for the set WB).



An exposure of -1.5 EV eliminates the red clipping, and this exposure reduction is approximately equal to the red WB multiplier.



Further problems occur with softproofing and printing the image as discussed earlier in this thread. Softproofing is limited by the gamut of the monitor. If one decreases the yellow saturation and no change is shown on the monitor, this indicates that the monitor gamut has been exceeded (LR 4 does have a monitor gamut warning).

The gamuts of the image, monitor, and printer can be examined with Colorthink and are shown below. The monitor is the NEC PA241W which covers most of the AdobeRGB gamut. The solid gamut is that of my printer, the wireframe that of my monitor, and the dots indicate the yellows in the image. The gamut of the image exceeds both that of the monitor and printer for yellow. However, since the monitor gamut exceeds that of the printer for the yellow, soft proofing should give good results.



Regards,

Bill



Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 8990



WWW
« Reply #16 on: April 25, 2013, 08:58:11 AM »
ReplyReply

What's the difference between doing that and reducing saturation in one or more colors in the HSL panel?

It's real useful to know if the data is really there, the 'blocking' is a result of the display, not the data. By soft proofing, one can get a better idea if the out of gamut 'issues' are due to the data itself (which then would require affecting situation) or just a visibility issue with the display and colors it can't reproduce. Kind of like seeing banding on a prevew. Is it in the data or a result of the display? Useful to know before you treat the problem.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
elied
Sr. Member
****
Offline Offline

Posts: 262


« Reply #17 on: April 25, 2013, 09:52:58 AM »
ReplyReply

Quote
It's real useful to know if the data is really there,
Yep, it sure is. Which is why Raw Digger is so great. As was Rawnalyze before it. Udi Fuch's UFRaw also had a Raw histogram (I suppose it still does, I haven't looked at it in several years). But these are all freeware or donationware, essentially projects by enthusiastic (and generous) members of the photographic community. Why does a slick commercial product like ACR/LR, the product of a well staffed (and presumably well paid) design staff and which describes itself as a comprehensive tool for the photographer, not have an input histogram as well as the output predictive histogram? Or input + gamma correction. Numerous photographers have ranted for years (justifiably) that the cameras should have one, but failing that, why not our Raw processors?
« Last Edit: April 25, 2013, 09:59:45 AM by elied » Logged

Roll over Ed Weston,
Tell Ansel Adams the news
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3579


« Reply #18 on: April 25, 2013, 10:57:55 AM »
ReplyReply

Yep, it sure is. Which is why Raw Digger is so great. As was Rawnalyze before it. Udi Fuch's UFRaw also had a Raw histogram (I suppose it still does, I haven't looked at it in several years).

Hi,

RawTherapee also has an optional Raw histogram, and it allows to adjust the WhiteBalance induced clipping before it happens.

Quote
But these are all freeware or donationware, essentially projects by enthusiastic (and generous) members of the photographic community. Why does a slick commercial product like ACR/LR, the product of a well staffed (and presumably well paid) design staff and which describes itself as a comprehensive tool for the photographer, not have an input histogram as well as the output predictive histogram?

Good question. Maybe they deemed it to geeky for the majority of their target audience?

Cheers,
Bart
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2783



« Reply #19 on: April 25, 2013, 10:59:26 AM »
ReplyReply

Yep, it sure is. Which is why Raw Digger is so great. As was Rawnalyze before it. Udi Fuch's UFRaw also had a Raw histogram (I suppose it still does, I haven't looked at it in several years). But these are all freeware or donationware, essentially projects by enthusiastic (and generous) members of the photographic community. Why does a slick commercial product like ACR/LR, the product of a well staffed (and presumably well paid) design staff and which describes itself as a comprehensive tool for the photographer, not have an input histogram as well as the output predictive histogram? Or input + gamma correction. Numerous photographers have ranted for years (justifiably) that the cameras should have one, but failing that, why not our Raw processors?

Excellent points. RawTherapee also has raw histograms as well as sophisticated highlight recovery and sharpening algorithms using Richardson-Lucy deconvolution.  LR/ACR are still my default processors, but I wish Eric and Thomas would include some of the requested features. Otherwise, they may suffer defections.

RT luminance and RGB histograms


RT Raw Histogram


Bill
Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad