PDA

View Full Version : Colour cast in E5!!


Kiwi Paul
15th December 2010, 07:46 PM
Here's a couple of photos taken with the E5 (first pic)and the E3 (2nd pic) at ISO 3200. It was taken inside under an energy saver light bulb. Both cameras had their white balance set using the white balance preset method, the E5 W/B is 2650 and the E3 2600 so no real difference, the exposures are identical (well the E5 used 1/30 sec and the E3 1/20 everything else is the same). These are RAW unaltered and converted to jpg using sRGB colour space with LR3.3. The E3 accurately shows the colours and tones, (incredibly so in fact) where as the E5 looks like I've selected "vibrant" mode or something, the colours look "garish" and are incorrect, the dog has an almost orange hue, the red and greens look too vibrant and the carpet has a red hue.

This is the second time I've tried this and even after resetting the W/B on both cameras the results are the same.

Has anyone else noticed anything like this, any suggestions?

All camera settings are identical and remember these are RAW not jpg's.

http://farm6.static.flickr.com/5088/5264351218_a979c64595_b.jpg http://farm6.static.flickr.com/5081/5264351996_2eedec5c91_b.jpg

Paul

Jim Ford
15th December 2010, 08:06 PM
They look pretty near identical on my calibrated CRT monitor!

Jim

Kees
15th December 2010, 08:11 PM
Too much red in the E-5 picture.

First you say you did conversion to JPG in LR en then you say that these are RAW?????

Kiwi Paul
15th December 2010, 08:25 PM
They are RAW pics converted directly to jpg (so they be posted on the net) via LR3 with no post processing.

Paul

Dogcow
15th December 2010, 08:28 PM
Just to make sure that it is not an Adobe RAW converter issue: do you get the same differences when developing the Pics in Oly Viewer 2?

Zuiko
15th December 2010, 08:53 PM
I/30th v 1/20th will give a more saturated (vivid) image but it doesn't explain the colour cast.

shenstone
15th December 2010, 09:20 PM
Too much red in the E-5 picture.

First you say you did conversion to JPG in LR en then you say that these are RAW?????

Looks like the 1st has a little more green to me not red ?

I have no specific answer as I own neither of the cameras, but agree with John (Zuiko) that it would be worth looking at the image in a few pieces of software to try and work out whether this is a camera difference or a software difference.

It could be Adobe not the E5 as even unprocessed, the software is rendering the file in some ways. Olympus are famous in some software circles for making undocumented changes to their ORF formats that can catch developers out. I had an issue when I 1st had my E510 that required a software patch.

If you can make the ORF files available somewhere I can take a look in the non adobe software I have

Regards
Andy

Nick Temple-Fry
16th December 2010, 12:45 AM
Could it be related to the differences noted here http://e-group.uk.net/forum/showthread.php?t=11235 - High ISO E-5 comparison

Not a solution/explanation - just wondering if different strategy for managing high iso + sensor capability are interacting, especially with custom wb, might be exacerbated by Raw not being processed in Viewer.

Nick

Ross the fiddler
16th December 2010, 05:27 AM
I would prefer to see the comparisons from Oly Viewer 2 first, since it is all Olympus & the other software, as good as it is, still has to emulate the intended conversion of Olympus RAWs, albeit with more flexibility.

Maybe it would be better to compare the images with similar manual settings of each camera too.

Ian
16th December 2010, 09:01 AM
Yes, I feel that what we may simply be seeing here is that the default view for E-5 RAW files in LR3 is different to the default for E-3 files in LR3.

You could try comparing in-camera JPEGs instead, but the best comparison would be in Viewer 2 as has already been suggested.

Ian

photo_owl
16th December 2010, 09:14 AM
Could it be related to the differences noted here http://e-group.uk.net/forum/showthread.php?t=11235 - High ISO E-5 comparison

Not a solution/explanation - just wondering if different strategy for managing high iso + sensor capability are interacting, especially with custom wb, might be exacerbated by Raw not being processed in Viewer.

Nick

Nick,

Can't see how this could be a factor as it's a jpeg engine 'thing', which would be duplicated in Oly viewer rendering if the defaults were used.

I agree that the effect is similar ie if the exposures are matched on these and you de-saturate the E5 image slightly they get to look closer - however there is a clear colour temperature difference in the two images on top of the exposure difference. Small but distinct.

However this is a software rendering issue as presented, and needs to be much better presented if analysis is going to be useful.

Again one of the problems we face in using software is that the manufacturers are also torn between the various performance elements ie Adobe know that if ACR has a rep for producing poor high ISO chroma noise on the E3 they might just be tempted to duplicate the trick Oly's jpeg engine used in the E3 in their camera profile!

Ian
16th December 2010, 09:18 AM
I've lent my E-3 to John (The Music Man) while his is being serviced, so I can't do a back to back comparison, but I will see if I can see a difference between the E-5 and E-30 later.

Ian

photo_owl
16th December 2010, 09:21 AM
I've lent my E-3 to John (The Music Man) while his is being serviced, so I can't do a back to back comparison, but I will see if I can see a difference between the E-5 and E-30 later.

Ian

I will try and run a 3 v 30 test as well today Ian

PaulE
16th December 2010, 09:26 AM
As far as I remember a default ACR convertion from RAW gives colour casts / very poor colours for all the Olympus cameras I've used that are not present in either OOC Jpegs or Jpegs/tiffs generated from raw files using Olympus software. So unless you are comparing the cameras using OOC Jpegs or jpegs/tiffs generated from RAWs using Olympus Viewer then IMO you can't say that the E5 comes with a colour cast - ACR probably introduces a colour cast but I'll bet it's not there in the original RAW files.

Kiwi Paul
16th December 2010, 09:28 AM
Thanks for the responses so far.
The in-camera jpg's show the same difference.
On my monitor at home (which is calibrated) the dog tones on the E5 photo really do look horrible, too orange.
By adjusting the W/B and tint in LR3 I can get the tones closer but I found if I get the dog tones correct the reds and green look even more vibrant and the carpet still has a distinct red hue that the E3 photo doesn't.
I set the white balance on both cameras using the one touch white balance method where you use a sheet of white paper and take a shot of it while the camera is in W/B calibration mode, both cameras were set up the same one after the other and the E5 selected a W/B of 2650 and the E3 2600. Changing the W/B in pp so they are the same doesn't make any appreciable difference as they are already so close.
The E3 colour balance etc is spot on, the shot was taken from my PC chair so I can view the photos on screen and look at the actual scene to compare.

I don't ever expect to different models of camera to be exactly the same, if it was just a mild discrepancy I wouldn't even be concerned but the E5 colours under that lighting really aren't right.

Paul

Kiwi Paul
16th December 2010, 09:31 AM
I'll look at them with Olympus viewer 2 tonight and see how they look. But the in-camera jpg's had the same effect.

Paul

Ulfric M Douglas
16th December 2010, 09:36 AM
... inside under an energy saver light bulb. ... the E5 used 1/30 sec and the E3 1/20 ...
Just a thought : what if the lightbulb 'cycle' has a different spectrum at any moment (I don't know!) and the shutterspeed difference is changing the colour balance.

I see too much Amber in the first one. Someone else sees too much Red, someone else sees too much Green. What a quandry!

Possibly a related problem ;
http://www.mu-43.com/f42/whats-causing-these-colour-shifts-8011/
Different camera, similar sensor/processor, similar random colour cast.

Jim Ford
16th December 2010, 09:54 AM
I see too much Amber in the first one. Someone else sees too much Red, someone else sees too much Green.

It means that the difference is so slight!

As I stated earlier, they look pretty near identical to me - at least at first glance. The main difference is that the top one has had slightly less exposure. I guess I could import them into Photoshop and do spot checks on the colours - I might try it later.

Jim

Daveart
16th December 2010, 10:41 AM
Hi I have just done a quick auto tone in cs5, looks closer to me now IMHO

Dave

photo_owl
16th December 2010, 04:50 PM
the E5 colours under that lighting really aren't right.

Paul

I was going through this again prior to doing some control files for Ian ( 3 v 30 to round off 5 v 30 and back to 3 v 5) but this suddenly stopped me.

As I read it you are saying that the difference in colour cast is only under these lighting conditions?

If so it's pointless me trying to produce meaningful files on the one hand, and on the other you would seem to be straight back to exposure and WB settings on the shots shown (and WB isn't just colour temperature).

Put another way you would seem to be illustrating a difference in the way the two cameras have set their WB from the same methodology and with that source.

Ian
16th December 2010, 05:54 PM
I've now posted an E-5/E-30 comparison here:

http://e-group.uk.net/forum/showthread.php?t=12305

Ian

Kiwi Paul
16th December 2010, 06:59 PM
I think I've may gotten to the bottom of it, but unfortunately my test subject got bored and wandered off putting further testing on hold.

It seems to be due to the fact the E5 is underexposing by approx 2/3 -1 stop, I tried some shots tonight and over exposed the E5 by 1 stop and there was a big difference, much better colours, still a bit more saturated than the E3 but much better. Altering the under exposed photos in LR3 didn't alter the colours that much so it seems correct exposure is crucial for correct tones with the E5 (well under this type of lighting anyway).
I only got one sequence fired off before Roz decided she was sick of being an unpaid and undervalued glamour model with her pictures been broadcast for the world to see and went on strike, so once I've renegotiated her terms and conditions (walkies) further analysis can proceed.

The OOC jpg's show the same characteristics too.

Paul

Dogcow
16th December 2010, 07:40 PM
That is very well possible.

The explanation is here (http://www.dxomark.com/index.php/en/Camera-Sensor/Compare-sensors/%28appareil1%29/682|0/%28appareil2%29/220|0/%28onglet%29/0/%28brand%29/Olympus/%28brand2%29/Olympus), it makes completely sense when you compare the ISO curves.

Kiwi Paul
17th December 2010, 09:06 AM
I played about last night and found the E5 under exposes by about 0.5 - 1 stop and over exposing the camera by those values improves things, still not perfect but much better.
It seems to over saturate the colours slightly too.
The in camera jpg's look better now, I set the contrast and saturation to -1 to calm things down a bit.

I can set up a profile in LR3 that suits my taste for the E5.

Thanks for all the comments and help.

Paul

snaarman
17th December 2010, 10:14 AM
I played about last night and found the E5 under exposes by about 0.5 - 1 stop and over exposing the camera by those values improves things, still not perfect but much better.
Paul

Interestingly, I get better results from my E620 since I set the global exposure bias to +3/6th EV. So (in centre weighted metering) my cameras nominally overexpose by that much in addition to any temporary exposure bias I dial in from shot to shot.

Until I did this I found I was lifting each shot slightly in Photoshop, and I was disappointed by the noise content. This lead me to my original conclusion that the IQ out of the E510 was better than the E620.

Now I have this permanent tweak in the E620 I would say the IQ is about the same. Maybe the ISO rating of the cameras is a tad optimistic?

Pete

photo_owl
17th December 2010, 10:16 AM
glad you are making progress

now we just need to deal with the exposure issue! I initially thought the E-P1 was 'underexposing', and to some degree it does to my eye. I've seen a number of E5 shots in lower light that look underexposed overall and the only thing they have in common is a small number of bright lights. The E5 seems to try and preserve all and any bright sources if it can do so whilst still providing an acceptable exposure at the other end.

I wonder if it's trying to be too clever.

Kiwi Paul
17th December 2010, 10:29 AM
Interestingly, I get better results from my E620 since I set the global exposure bias to +3/6th EV. So (in centre weighted metering) my cameras nominally overexpose by that much in addition to any temporary exposure bias I dial in from shot to shot.

Until I did this I found I was lifting each shot slightly in Photoshop, and I was disappointed by the noise content. This lead me to my original conclusion that the IQ out of the E510 was better than the E620.

Now I have this permanent tweak in the E620 I would say the IQ is about the same. Maybe the ISO rating of the cameras is a tad optimistic?

Pete

Yeah the DXo ISO sensitivity show the E5 to be around 2/3 stop optimistic and I would agree with that from my recent testing (well at 3200 anyway).

How do you adjust the global exposure bias?

Paul

snaarman
17th December 2010, 10:34 AM
Yeah the DXo ISO sensitivity show the E5 to be around 2/3 stop optimistic and I would agree with that from my recent testing (well at 3200 anyway).

How do you adjust the global exposure bias?

Paul

Ooo-er. IIRC its buried in one of the spanner menus on the E620.. Let me check my computer:

Spanner "I"
Exposure Shift (on the E620).

You can dial in separate exposure shifts for ESP, CW and Spot metering modes. As an aside, the E620 is enormously tweakable, and I guess the E5 is also. So much so that I made myself an Excel spreadsheet with my 106 settings so I can find them "just in case"

Pete

Ross the fiddler
17th December 2010, 01:27 PM
Ooo-er. IIRC its buried in one of the spanner menus on the E620.. Let me check my computer:

Spanner "I"
Exposure Shift (on the E620).

You can dial in separate exposure shifts for ESP, CW and Spot metering modes. As an aside, the E620 is enormously tweakable, and I guess the E5 is also. So much so that I made myself an Excel spreadsheet with my 106 settings so I can find them "just in case"

Pete

For the E5 it has been shifted to adj J but still called Utility & all 3 metering modes can be shifted seperately. I remember Wrotniak chose to have -1/6 for his spot metering on the E30 (which he had set for AEL).