The Holding Tank

Home Page Photography

Photography RSS 2.0 Feed

| Photos & Themes RSS | Hardware | Technical | Links | Books and Magazines  | Downloads |

Technical

| Flash Tests | RAW Converter Tests | Depth of Field, Image Quality and Sensor Size |

| MTF | Resolution | Cropping and Printing | Image Stabilization and Magnification |

| Lens Image Simulation | Anti-Alias Filter and Sharpening | | Tilt and Shift Lenses |

Lens Distortion Correction and Image Detail | Digital Darkroom Workflow | Optical Coatings |

Multipass Sharpening Comparison

TOC

Introduction

Capture Sharpening

Image Content Sharpening

Output Sharpening

Using Capture Sharpening

Using Capture and Image Content Sharpening

Single Pass Comparison

Evaluation

Inkjet to Halftone Delta Sharpen

Conclusion

Introduction

This short note looks at various multipass sharpening workflows. It draws on the following resources:

TRL:         TLR Sharpening Toolkit (Action Based Toolkit)

TRLP:      TLR Professional Sharpening Toolkit (Script Based Toolkit)

Fraser:     Real World Sharpening with Adobe Photoshop CS2 -  Bruce Fraser (Book)

Canon:    Canon Document EOS_Digital.pdf (Capture Phase Only re p4)

Some of the processes have been implemented in Photoshop actions downloadable here.

A fairly complete set of actions based on Fraser's book are downloadable from here.

The below details the basic processes with the baseline parameters for each.

Note [1]: Frasers Image Content sharpening is an edge mask protected USM sharpening tuned to the dominant image information frequency. As such it seems to me to belong with Creative Sharpening although Fraser lumps it with Source (Capture) Sharpening. Fraser is not completely consistent in the use of Image Content Sharpening. On p160 he states that Source and Image sharpening can often be combined. However in other parts of the book (most notably p263 et seq) he performs these operations separately. It seems his normal practice is to perform source sharpening as an unmasked operation, then mask they layer as required for the content sharpening and apply the image sharpening to that layer. It is important to do this otherwise the source sharpening is unprotected by a layer mask although this point is not made in the book. 

Note [2]: If required the first stage should also include a layer to perform noise reduction, normally with a surface mask. Fraser recommends performing NR before Source (Capture) sharpening. This seems like good advice however I often find the same sharpness with lower noise can be achieved by following the capture sharpening with noise reduction. NR is not investigated in this note.

Note [3]: The creative phase can involve several layers of differing sharpening and blur applied typically by a paint mask with a luminosity blend mode. This is not investigated in this note.

Note [4]: Note that Fraser varies some of the guassian blur and calculations methods for layer mask creation in different presentations on the same process, so the baseline parameters should probably be taken as a starting point.

In the following sections a photo with several different textures is used to evaluate the different sharpenings. This is reproduced below with two 100% crop frames showing the crop area for the native resolution (smallest crop area) and the output (print) resolution (largest crop area).

The whole photo is reproduce below.

The exposure information for this shot is EOS 20D 17-40f4L @ 40mm ISO 100 f10 0.3 Sec Tripod. RAW conversion involved minimal correction for white balance and highlight level and output to16 bit PSD, and for simplicity in this exercise the sRGB colour space.

Stage 1:Capture Sharpening

100% crops at native resolution are shown below for the various capture sharpening default settings.

All the sharpening methods seem to provide a similar degree of apparent sharpening. Most obviously standing out is the Canon recommendation that, due to no use of layer blend if, fails to protect the highlights in the moss.

The most complex in terms of processing is the TRL which include a complex edge mask, however this will provide better noise suppression for areas poor in detail in noisy shots.

Note that earlier modeling based on typical characteristics of camera anti-alias filters suggests that the Canon recommendation should provide a neutral level of sharpening in frequency response terms. As all these methods have similar strengths that conclusion can be extended to all of them.

Stage 2: Image Content Sharpening

This is a Fraser process only and is effectively an automatic creative sharpening.

Fraser defines three different processes for three different spatial frequencies, Low Frequency (LF), Medium Frequency (MF) and High Frequency (HF).

These are in turn added to Fraser Source (capture) sharpening and 100% crops shown below.

Strangely the HF sharpening is not the strongest sharpening, the most effective seems to be the MF.

Stage 4: Output Sharpening

Using Capture Sharpening

Only Fraser and the TRL Pro toolkit provide suggested sharpening for Half Tone. Below the examples are TRL capture sharpened, flattened and resized to 5.2X4 inch at 300 dpi using Bicubic Sharper (my normal procedure and also recommended by Fraser).

The baseline output sharpening is suggested for half tone 300dpi with a 150lpi screen on coated paper. The 100% crops are provided below for comparison.

Here we can see Fraser's baseline recommendation is much stronger.

For Comparison here are the recommendations for 300dpi Inkjet printer on Glossy paper.

Again the Fraser sharpening is stronger. However the HT sharpening recommended by Fraser is differently stronger compared to the TRLP than for the Inkjet case. Note the Fraser Inkjet sharpening is a combined USM and Highpass and quite different in execution from the halftone which is a simple USM sharpen.

Using Capture and Image Content Sharpening

Working now with the Fraser Source and MF Image Content sharpening layers, the effect of the Fraser Inkjet and Halftone sharpenings is investigated.

As can be seen the Image Content is adding considerably to the overall sharpness even when the output sharpening is added.

Single Pass Comparison

Here the RAW conversion has been simply scaled using Bicubic Sharper to the print resolution and size and the output sharpening applied. There are two sets of crops, one with the default layer luminosity and another with the same process but the layer luminosity set to 100%.

Although increasing the sharpening layer luminosity improves the overall strength of the sharpening, it can be seen the multi pass procedure yields better fine detail. 

It seems unlikely that a single pass will be as effective as multi pass as the multi pass permits server breakpoint in the modified spatial frequency response.

Evaluation

Capture and Output

The only way to evaluate these output sharpenings is to print them. Not having a halftone printer I have just printed on my desktop inkjet (Epson Photo 1290 on premium Glossy paper). A composite 10X8 file of all four outputs from the Capture and output sequence was created and printed.

Interestingly the differences between all four sharpening methods are not as large as might be expected from the 100% crops.

At normal viewing distances both the HT cases look much lighter around the moss due to the more intense sharpening halos as compared to the IJ cases.

Close examination under a loop shows the moss to look crunchy in the Fraser HT case compared to the TRLP HT example. Differences between the two on other textures are less obvious between the two.

In terms of the inkjet output, I would say the Fraser example is clearly oversharpened, but then it is intended for halftone printing.

Capture and Image Content plus Output 1

Another 10X8 composite this time containing the Capture + Image Content + Output Sharpening for both Halftone and Inkjet are the top pair. Along with this are the bottom pair containing the resized RAW sharpened only with the two output sharpenings at the default layer luminosity were printed and evaluated.

Capture and Image Content plus Output 2

Another 10X8 composite this time containing the Capture + Image Content + Output Sharpening for both Halftone and Inkjet are the top pair. Along with this are the bottom pair containing the resized RAW sharpened only with the two output sharpenings at 100%  layer luminosity were printed and evaluated.

Inkjet to Halftone Delta Sharpen

The question of converting a file intended for inkjet output to an approximation of that required for halftone output arises. To evaluate this concept the Fraser file for Source, MF Image Content and 300 dpi glossy inkjet Output is evaluated.

Here the inkjet file (far right) has had an additional sharpening layer (centre) to provide a similar sharpening level to the halftone output file (left).

The sharpening layer used was the Fraser 300 dpi 150 lpi glossy sharpening process, but the layer opacity was adjusted from the default 66% to 20%.

This concept seems feasible although further optimization is required. Additionally how well this process would work with files that have been sharpened with USM unmitigated with layer blend if needs further study.

Conclusion

People could do worse than to target sharpening for their inkjet printer at the target print size and resolution for HT printing although this would be slightly under sharpened assuming the TRLP and Fraser baseline suggestions are good.

The only way to know for sure is to print on the target HT process.

The Image Content can make a significant difference to the overall image sharpness. It seems likely that Fraser assumes that Source and Image Content will be added as a minimum to make a user neutral file, requiring only resizing and output sharpening. 

It seems unlikely that single pass sharpening can replicate the quality possible from the multi pass process, however if the ratio between natural and output resolutions is sufficient a single pass approach may be an acceptable compromise.

It also seems there may be scope to delta sharpen between files prepared for inkjet output to a file that can be printed with reasonable quality on a halftone process.

Last Updated 05/06/2008

All Content © 2005-09 Lester Wareham All Rights Reserved     

All material is supplied  as is and without warranty, use at your own risk.  

All opinions stated are the authors own.

All quoted information remains the copyright of the respective authors.

Contact: (anti spam remove underscores)

Note this email gets spammed a lot so if you want to get your message through please prefix "Holding Tank" in the title

f_ w_ a_ r_ e_ 9_ 0_ @_l_i_n_e_o_n_e_._n_e_t