• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname

    Register

    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

Will Silkypix 3 be a major contender in RAW conversion?

John_Nevill

New member
Some of us may have heard of Ichikawa Soft Laboratory Co (ISLC), others may have not.

I've been a advocate of Silkypix 2 for about 8 months and more so since the demise of Pixmantec's RSP.

Silkypix 2 was received with mixed opinions, some like its colours and rendering (for portrait work), while others see it as a tad "plastic".

I personally like its flexibility, rich feature set and controllability, however the user interface is, well a little strange to say the least!

Last week I received an email from ISLC (in Japanese) outlining beta 3. A few clicks using Babelfish and a host of new Silkypix features jumped off the page together with a redesigned user interface.

The new user interface seems to be more akin to the C1/RSP style with a thumbnail filmstrip, main screen and editing tools.

beta_gui.jpg


If you are an existing user of Silkypix 2, you will know its feature set, for the uninitiated it includes pretty much everything that C1 or RSP has to offer (except RSP’s highlight / shadow / vibrancy) with the added functionality of CA removal, lens correction and some serious parameter tweakability!

Silkypix’s NR, highlight and colour adjustments are very controllable indeed (you may well need a degree in physics to understand some of them).

The earlier user interface provided floating menus and was great for spreading across two monitors. I believe this also continues in v3.

So what does Silkypix 3 have in store, well a quick peak at the revised (and translated) feature list identifies:

a) New User Interface
b) Faster rendering engine
c) Updated Noise reduction system with noise cancellation
d) SilkyTaste (pretty much translates to Picture Styles)
e) A magnification CA removal tool with colour mode and presets
f) Updated model sharpness with detail and contour emphasis
g) Updated Lens correction
h) Independent user defined presets for most functions
i) Improved batch processing
j) ICC profiles - I need to validate the extent of this!
k) ICC printing to include contact sheets
l) Jpeg, TIFF and RAW viewing / editing ability
m) Support for a myriad of RAW files and additional functionality for Fuji S3 cameras.
n) Mac and PC platform

I've since emailed their support team and asked for a release date for the English version. It’s definitely on the cards but I was met with a cautious "when it’s ready" reply. Dare I speculate this as being early 2007?

So do we have another major RAW contender?

Well, it may not be Lightroom in the context of DAM, but I believe it could be a welcome RSP replacement.

I must say that the speed at which new camera support arrives is pretty impressive; I do wonder whether ISLC have better business collaborative arrangements than others.

The Japanese only retail price is quoted at 16000 yen (£72 or $138), while the upgrade price is quoted at 8000 yen (£36 or $69).

For the record, I am in no way connected to Ichikawa Soft Laboratory Co, I just wanted to pass on my findings and initial perceptions.

I also believe that it’s great to see some independent competition building again, especially with the imminent and welcome release of DX0 4.
 

Ray West

New member
Hi John,

I downloaded silkypix some time ago, (v2) and left it in free mode. Most of its capabilities were disabled. I've just activated the 14 day free trial, having read your thread wrt v3, to play with it some more. I find it a bit slow in making any changes. Other sw I've tried is either much faster, or there is a reduced preview area that is manipulated, before the changes are applied to the whole image. How are you using v2, do I have to rtfm? is there a preview mode or similar?

Best wishes,

Ray
 

Asher Kelman

OPF Owner/Editor-in-Chief
John,

How intuitive is it to follow the menu from the Japanese?

It would be nice to have a screen capture marked up with the basic stuff in English!

Anyone game for this?

Who knows Japanese?

Asher
 

David Robertson

New member
Hi John,

Thanks for the "heads up" on Silkypix 3. Its great news.

I've only just started using Silkypix and am stunned by the colour it offers. Its probably my lack of knowledge and skill, but I have never enjoyed RAW processing with either C1 or RSP, despite their workflow advantages. Processing images with thoughts of "that will have to do" was common. Now, with Silkypix, I'm producing images that match what I envisaged when I fired the shutter.

Cheers

Dave
 
I have been using the Silkypix beta versions since May, step by step they have implemented new functions and now they are close to releasing the Silkypix 3 to the Japanese market.
My findings so far:
Improvements compared to SP 2 IMHO:
- The filmstrip with detailed preview speeds up the sorting of your pictures.
- Highlight compression feature (This made me use the Japanese betas instead of SP2)
- New better sharpening method (the old method can still be used).
- Semi automatic lateral chromatic aberation adjustment.
Missing features/ less good
- The highlight handling is still problematic, if you have blown out highlights you can get some strange color effects (=ACR)
- Color space for output is still limited to sRGB and Adobe RGB
- The user interface is basically the same as in SP2 and might give problem for beginners, you will find the basic functions using the presets but if you want to do some fine-tuning you can get lost among the extreme number of functions.
Good functionality kept from SP2
- Color handling/color adjustment features/a number of color profiles for each camera
- DNG/Linear DNG import (camera profiles for DNG import)

Since my Japanese skills are limited a might have missed some things and might be complaining about features that's already implemented.

Overall I'm kind of 4 times disappointed at the moment , RSP will die, LR beta 3 for W, SP 3 and DxO V4 beta. I had hoped for more but maybe one of the sharp versions will come out as a winner.

BR/ Stefan Hellstrom
 

Asher Kelman

OPF Owner/Editor-in-Chief
Maybe the solution at the moment to the language issue is for users to make available screen shots with annotations of what they know and new posts could fill up the blanks?

Anyone want to start?

Asher
 
The basics functions in SP3 are the same as in SP2 so the english html manual for SP2 or using SP2 will get you started.
For the SP3 you also have a html Japanese manual (4,5Mb), if someone can host it on the web it can be viewed using the Babel Fish translator.
I will change my internet provider next week so it's not much use for me putting it out.

I have done some more testing with partially blown highlight pictures and it seems like the SP3 beta performance is worse at default settings than the SP2 for some reasons. The SP2 is about equal to ACR, RSP or DxO comes out on top in my opinion .
BR/ Stefan Hellstrom
 

Dierk Haasis

pro member
I tested Silkypix some time back (yes, 2.x). The two main problems have to be tackled by the developers, not by their paying customers:

1. Translation - the current on-line translation tool approach does not work.
2. UI - flexibility and sheer power is good, as long as the user has no problem finding his way.

Most of us - all, I guess - look for ease of use to enable quick processing [of any kind], so we cann minimise our comp time and maximise our photo time. Silkypix is far out on these.

Personally I am not inclined to learn Japanese for the next five years just to have another converter lying around.
 

John_Nevill

New member
Dierk, I'm not sure that I follow you on the translation issue, v2.x comes with an "engrish" manual in html form and includes everthing from shortcuts to function explanation.

Ray, as for speed, on my duo core latpop it performs a tad slower than RSP, but its not that significant to worry about. Have you tried DX04 or LW beta, now they are slow by comparision.

Stefan, I agree with you on the blown highlights issue on WDR images, i'm wondering whether this more to do with the low res LUTs of the embedded ICC profiles. I really hope they provide user ICC profiles. But looking the way the RAW market is going only C1 continues to offer these.

Its real shame as I have always created and used my own camera ICC profiles for Hi sat, Lo sat, WDR, emulation and B/W in RSP. Generic profiles just don't seem to cut it for me.
 

Dierk Haasis

pro member
John_Nevill said:
Dierk, I'm not sure that I follow you on the translation issue [...]

Two times I looked into Silkypix, both times it practically threw me away by its very bad translation. I am talking not about style but simple wording => usability. Since I don't know Japanese I can freely say: they urgently need to get a professional translator, most likely a native speaker of English (at least somebody with a good knowledge of the correct industrialect©.
 

Dierk Haasis

pro member
Strange, mine [Suzuki Savage, Honda CB 500, Honda Fireblade 1999, Honda Fireblade 2002] always came with quite understandable German manuals - although I only needed it for the last bike to change the cockpit settings.
 

John_Nevill

New member
Dierk, their english manuals use to be awful, although things have improved with the O&M for my current bike (Suzuki TL1000S)
 
Silkypix 3 Beta Highlight recovery

I have done some more checking on my problems with the highligts for some pictures developed with SP 3 Beta.
-This is the picture at default SP 3 settings:
http://static.flickr.com/96/238570830_dedc4da242_o.jpg
- The second one is using the Higlight Recovery slider set to 3, everything else at default:
http://static.flickr.com/98/238570831_9bd9ec6b2c_o.jpg
Some nasty artifacts shows up in the highlight area of the girl's face.

- The third one is the same settings except for that the saturation-tone and the Restoration slider are set to zero:
http://static.flickr.com/89/238570833_913f34ca56_o.jpg
Number four has the exposure slider set to -0.9, one snap lower contrast and increased gamma:
http://static.flickr.com/95/238570834_7cfcc99eaa_o.jpg

If you have a perfectly exposed picture and add +1 on the exposure slider to lighten the shadows when the Highlight recovery slider will work without problems.

/Stefan Hellstrom

Edited for privacy
 
Last edited by a moderator:

John_Nevill

New member
Stefan,

Interesting, that looks very much like bad ICC profiles, if the LUTs are poorly constructed, you get posterization and blotchy colour artifacts as the RGB values begin to clip in an irregular way.

I've downloaded the beta 3, but my Japanese is non existent, so I can't tell if you can use your own ICC profiles or not. This will be the deciding factor for me.

Thanks for the heads up on this!
 
Hi John,
I dont think it's a profile thing, it's more like that the highlight retention function needs manual support from adjustment of the tone and restoration sliders to reach the same level as other RAW converters.
This is the highlight controller panel from SP2:
238773319_3f906b72cf_o.jpg


And this is the same panel in SP 3 beta with the new highligt compression slider at the bottom:
238773320_f4a37efad8_o.jpg


I think the highlight features in other converters is doing some behind the scene magic to limit artifacts. Getting some color on a 255,255,255 area must be a looking at the Neighbour area and guessing in a smart way.
About the profiles: The beta manual clearly states that you have sRGB and AdobeRGB as choises for working space so it's the same as SP2. The color gamut for a Canon DSLR a lot wider than AdobeRGB so I can't understand why one of the most feature rich RAW converter in the world is not allowing the use of wider profiles.

/Stefan Hellstrom
 

Denis de Gannes

New member
SilkyPix user one year.

I purchased SilkyPix English V2 11 months ago while awaiting the release of RSP.
RSP became my converter of choice because of its better (imho) work flow and speed. However I returned to it many times when I needed to do large 13 X 19 and 16 X 20 prints from my E300 ORF files. I find the color, tonality to be an improvement over RSP.
I have recently been testing the Lightroom Beta 3 and am hard pressed there to better the SilkyPix conversions. The SilkyPix team seem to be extremely well organised and most times is in the lead in providing support for new camera models. I have always believed they have an inside track with the Japanese camera manufacturers. Panasonic has begun providing SilkyPix as the packaged RAW converter with the new models.
I would certainly upgrade when v3 becomes available in English.
 

Asher Kelman

OPF Owner/Editor-in-Chief
Denis, could you post pairs of images processed in SP and LR or other processor.

Also would you or anyone else like to contribute a RAW file for us to have a standard library of reference files to test with various processors.

Asher
 

John_Nevill

New member
Stefan, I'm hoping for custom input profiles for cameras not output colour spaces such as AdobeRGB and sRGB.

Denis, its good to see you over here, I frequently read your posts on the pixmantec forum!

Asher, would an IT8 target RAW from a 1DN be useful?. I also have a nice lake district landscape from a 20D i'd be willing to donate.
 

Asher Kelman

OPF Owner/Editor-in-Chief
John,

That is so perfect!

The target came from where? It could go to be measured with a colorimeter?

Then we would have an interesting standard.

We need about 20 different RAW files to start with. Then we'll make them available for download. Once we recognize the issues with each one, we will know what to look for.

We could also have an unmodified DNG or postion of a DNG. But what processors can uses DNG files?

John, how would you like to be the one getting this together. Start a thread and email to anyone who might be able to contribute? What do you think? If you don't have the time, I understand! :)

Asher
 
Last edited:

Dierk Haasis

pro member
Anybody nice enough to explain to me on what basis we should judge different renditions of RAW converters?

Before anyone runs into this knife I opened up, let me say that in the early days of converters a few years back, when there were very visible differences and relatively few programs actually capabel of tweaking, I ran such a test. And I did it it over the course of the past 18 months, this time privately and focussing on something very different to the first test.

For the moment I will set aside the purely subjective argument behind which converter produces the best colours, let's assume it is possible to judge on an objective, scientific basis. We have to define the limits of the test very clearly or we are unable to judge at all. What does that mean for the converters? Do we use the default settings, which ones, those from the camera [Nikon Capture wins hands down], those the programs' developers deem a good strating point?

Then, how do we get the same input? The only way this can be assured would be that one of us with loads of different cameras - and several bodies for every model - and a tested, comparable lens for the different makes shoots one and the same target several times. But then, how to decide which photo to use? Putting up a target for printing is so obviuosly error-prone, we shouldn't even think about it.

Leaves the question of aim - are we testing for print output, on an offset machine or a capabel SOHO printer? Since our comparison is only feasible by using the Internet, sRGB'ed resized versions would be our goal, wouldn't it?

QM2_blue.jpg
QM2_orange.jpg

Which one is correct - and does it matter?

Eventually it comes down to personal preference - curiously not so much in how any one program handles the colous but personal preference of the UI and work-flow. In the best of cases the default settings are so good that we rarely need to change anything [WhiBal used*]; with ACR we have to tweak and fiddle a lot.




*For this plug I really should get something ...
 

John_Nevill

New member
Asher, No problem at all, give me a few days as my new host provider has had technical diffiulties which has resulted in total loss of my web sites.

I may also be able to allocate ~500mb of web space to house the RAWs. Leave it with me and i'll PM you with the details

My host loss couldn't have come at a worse time!, I have National Geographic interested in one of my images for the front cover of their kids magazine and I couldn't email them the proof. This is my first big opportunity, I really cant afford to let it drift.
 

Asher Kelman

OPF Owner/Editor-in-Chief
John,

I hope you can recover everything!

Dierk Haasis said:
Anybody nice enough to explain to me on what basis we should judge different renditions of RAW converters?................

Introduction: Image processing software has the capability of producing an array of results that depend on

a) Image characteristics (the source of the image, nature of the subject, textures, lighting, contrast, dynamic range

b) Capabilities of software programs

c) Person doing the image processing (taste, imagination, time available, skill and more)


Proposed purpose

1. To have a library of RAW/DNG/JPG "as shot" images from various cameras representing a variety of scenes and inherent processing issues. This would be for OPF members to test as they wish with whatever processor they can use.

2. To Recognize in our collection of images, particular problem areas in processing

3. To allow processed images to be shared among OPF members so different versions can be printed by individual members on their own computers for personal evaluation.

4. To share personal techniques, experience and opinions on different processors.

5, to learn how to overcome particular processing issues by sharing challenging RAW files.

The final end points are the personal preferences of users.

Dierk Haasis said:
Do we use the default settings, which ones, those from the camera [Nikon Capture wins hands down], those the programs' developers deem a good starting point?
No, what ever you choose to be "best" in your own judgment. However, each would want to have the setting chosen to give a reasonable idea of how to repeat that processing.

Dierk Haasis said:
Then, how do we get the same input?
We don't, that's the idea. We'll start with the same images. However, hopefully, we'll end up with different files and this will yield different prints, some of which might tell us something new. One look on your screen will tell you whether for you it is relevant to test print the various processed versions posted.

Dierk Haasis said:
But then, how to decide which photo to use?


You decide by submitting a RAW, DNG or jpg, as shot, that you feel would be worthwhile to optimize.

I would collect the images with the features said to be important for each image, and we'll see how to choose from that point. It depends on the number of images received.

Dierk Haasis said:
Leaves the question of aim - are we testing for print output, on an offset machine or a capable SOHO printer?

Any printer you have available or feel it is worthwhile to pay for. You might print it on your Epson or at Costco’s Noritsu or Lightjet or send it to a fine printer. It all depends on how relevant, significant, well implemented and important it is for you.

Dierk Haasis said:
Since our comparison is only feasible by using the Internet, sRGB'ed resized versions would be our goal, wouldn't it?

Well sRGB would be posted here, of course, as we will host them and don't want to have bandwidth issues.

However, where there is interest we will request full size image file (with © attribution) for OPF members to download. These will be available and licensed to contributing OPF members to download to study and to test print only.

Dierk Haasis said:
personal preference of the UI and work-flow. In the best of cases the default settings are so good that we rarely need to change anything

This is one aspect that will come out. We all can't be equally involved time wise in every DIMP! However, we can perhaps learn from each other what to avoid and what may be a solution for particular issues.

Asher
 

Don Lashier

New member
Dierk Haasis said:
Anybody nice enough to explain to me on what basis we should judge different renditions of RAW converters?
My explorations revealed that most apparant differences in raw converters have to do either with WB interpretation or default tonality curve.

The significant differences (that can't be easily adjusted out) have more to do with noise handling and artifacting. eg the gory details, more gory details and high ISO noise. These are old studies but not much has changed.
Which one is correct - and does it matter?
This (Queen M2) appears to be primarily a difference in WB interpretation although obviously the tone curve is different also.

- DL
 

Dierk Haasis

pro member
Don Lashier said:
This (Queen M2) appears to be primarily a difference in WB interpretation although obviously the tone curve is different also.

Only the WB has been changed - and only I know which is right because only I saw the original scene. Now RAW converter or program can do anything testable when it comes to colour and tonal interpretation. Which was the whole point of my diatribe: Everybody taking a few hours off can produce the same result from one image in every [well regarded] RAW converter available.

Since only the manufacturers themselves know what their cameras do, only they can produce appropriate algorithms in software, hence, Nikon Capture [4 and NX] come up with very nice looking results from the start. Others even have to guess the white balance value of the scene! So everybody involved in testing RAW converters will have to decide on what grounds he judges them - and image quality may be a nice catchphrase but won't cut it.

As Don points out, there are a few areas in which it is possible to test for quality, mainly noise [reduction], perhaps sharpening. That's why my original test - which was highly controversial because of my decision - used the defaults of the converter. Not one dud among them, all winners with slightly different renderings, a bit cooler, a bit warmer, more saturated yellows ...*

Eventually it comes down to time used to achieve the same result in all converters tested: Which one is the easiest and quickest to get me the result I want?



*BreezeBrowser, QImage, ACR 2, Capture 3 and 4 (I think I had one more but I forgot which.)
 
John_Nevill said:
Stefan, I'm hoping for custom input profiles for cameras not output colour spaces such as AdobeRGB and sRGB.

John, I can't see anything in the program or the manual that looks like an option for inputting custom profile. For my hobby use the number of canned profiles and the SP3 function for saving presets makes the need for custom profiles go away, but for someone doing art reproduction it might bee a different story.

BR/Stefan Hellstrom
 
Dierk Haasis said:
As Don points out, there are a few areas in which it is possible to test for quality, mainly noise [reduction], perhaps sharpening. That's why my original test - which was highly controversial because of my decision - used the defaults of the converter. Not one dud among them, all winners with slightly different renderings, a bit cooler, a bit warmer, more saturated yellows ...*

Hi Dierk,
Since I have been using DxO 3 to output linear DNG to SP2 to get the colors as I like them it gave me an idea:
For RAW converters that can read linear DNG you can compare default color rendering after demosaicing by using a digital target, you can also get an idea how the white ballance is handeld.
To produce the target you need a program like ACR, Lightroom or DxO that can output a linear DNG.
If you for example use a Bruce Lindbloom elctronic LAB version of GretagMacbeth ColorChecker to create an AdobeRGB linear DNG version and open it up in Silkypix using the default DNG profile and default settings you can notice some things:
-The WB is set to 5200 by SP, the neutral CC20 patch is slightly off target. Move the WB to 5000 and it will go neutral.
- The color rendering of the target will be totaly different compared to the electronic original (The default SP profile is not flat/neutral)

If you do the same thing with fex. Lightzone it will come up with WB set to 5000 and if you try to set the WB on the CC20 patch it will stay at 5000K. The on screen apperance seems to be identical to the electronic original.

If you use a number of converters to do their versions of the test target and gather the values for the color patches it might be used to do an overview of how much off they are and in what direction.
I'm not sure if this is a good idea or not but at least it can give you some idea of what is happening after the demosaicing.

BR/Stefan Hellström
 

Don Lashier

New member
Stefan Hellstrom said:
If you use a number of converters to do their versions of the test target and gather the values for the color patches it might be used to do an overview of how much off they are and in what direction.
I'm not sure if this is a good idea or not but at least it can give you some idea of what is happening after the demosaicing.

I don't think it would be very worthwhile. The reason is that converters apply a tone curve of some sort and this affects both color and saturation. IOW the RC providing the best target match might well be that one that produces the worst images, at least for standard purposes.

As someone mentioned earlier, this is a subjective matter, and the tools and workflow facility are a big part of the equation.

- DL
 
Silkypix 3 English Beta is out

The Silkypix 3 English beta is out from today, it can be downloaded and used for evaluation during 30 days.
Link to download page: http://www.isl.co.jp/SILKYPIX/english/sp_ds3/download/

Compared to SP2 I still like it a lot for portraits and some landscape pictures. My main problem area is still details in overexposed highlight areas.
I also retested a ISO 800 picture that I remembered caused problems for the semiautomatic noisereduction in SP2 (the darker areas showed some banding effects with default NR). SP3 seems to do better with this image. The picture also has a lot of local highlight areas where SP is having problems with the details (fex the clock on the left side)
Just for fun to compare the starting point I also run the same image in RSP, Lightroom B4 and DxO.

SP3 B3 E default settings: http://www.lsn.se/apap/OPF/IMG_2713_SP_B3.jpg

DxO V4 default settings:
http://www.lsn.se/apap/OPF/IMG_2713_DxO_V4.jpg

LR B4 W default + auto ajust tonality:
http://www.lsn.se/apap/OPF/IMG_2713_20051126_LRb4.jpg

RSP CE default + auto adjust (alt E)
http://www.lsn.se/apap/OPF/IMG_2713_20051126_LRb4.jpg


BR/ Stefan Hellstrom
 

John_Nevill

New member
Stefan,
I got Silkypix beta 3 yesterday and had a play. I ran a few images through it last night and compared it against LRWb4. I think LRWb4 has the edge on IQ, but there's some nice features buried in the toolset.

The highlight controller has some presets for highlight dampening, this did a good job on sunsets etc. (have you tried it?)

The CA tool has a picker which calculates CA, just click the aberration and its gone.

I must confess that the sharpness preset terminology is a bit strange, but in the main most settings are now redefinable so one can call them what they like.

Did you notice that a lot of the tools adjust there default settings depending upon inage?

I suppose the the main benefit is that its very stable, fast and can be used confidently as a helper application to your DAM.

I didn't notice the banding issue (and will check it), but I might add that LRWb4 also exhibits some strange artifacts in highlights areas.

The biggest benefit for us is that we have a choice and the concept that one application will be the panacea for all our needs is long off.
 
Top