After culling any no hope frames all frames are renamed with a category, friendly title and perhaps additional info such as a map reference and a three digit sequential number fff. A date sequence is added to ensure unique names for photo servers and aid backwards location.

e.g.

Landscape Castle in the Rain SU567234 002 091005.CR2

For ACR sidecar files these are stored in the same directory.

Currently I store CR2 files and only save PSD files if significant Photoshop work is required, ie spotting, creative sharpening. In this case a mirrored directory structure is used.

…/ pictures/20d/yy/<title> dd-mm-yy

If the shot only requires ACR dynamic range, colour temperature and crop adjustments and subsequent default capture sharpening I don’t bother to save this as it can easily be reproduced. Also raw conversion algorithms are improving all the time.

A two stage archive process is used for photo files and general computer data.

The computer used has two drives. The Windows Backup Utility is run as an appending incremental backup by a script as a daily scheduled task, the script copies the archive to second drive so two independent copies are stored. Once a week the archive file is moved to a pair removable Firewire harddrives.

I use the Adobe Bridge colour tags as follows:

Red:        For RAW files used together to make a single image using layers or focus stack software.

Purple:    For RAW files where a secondary master PSD file exists containing a mixture of adjustment layers and possibly custom capture/creative sharpening or noise reduction layers.

Yellow:     Is used for PSD files that already include a capture sharpening layer. They may also include adjustment layers, creative sharpening layers and noise reduction layers.

Back to TOC

RAW Conversion

Although I found the Canon supplied DPP produced good results I have now switched to using Adobe CS2/ARC mostly because of the integrated solution.

I use RAW for the additional control and quality, particularly allowing me to set the white and black points precisely and to set the colour temperature and contrast or make curves adjustments without loosing bit resolution as one would with JPG.

I set the black and white points to about 20 min and 235 max per channel as indicated by the dropper tool in ACR to ensure some margin for the output device.

So, having set the major dynamic range and colour corrections and cropping as required I then open to 16 bit in the Adobe RGB colour space in Photoshop for further processing and save the RAW edits.

I have ACR sharpening set to preview only, as I do the sharpening in Photoshop using multi pass sharpening methods (see the next section). There are some examples of various sharpening workflows here.

I only use noise reduction for ISO 1600 and 3200, normally trying to confine this to colour noise filtering mostly to minimise loss of sharpness. Additionally I may use surface layer masked gaussian blur in Photoshop to provide luminance noise reduction. This is very flexible and controllable.

I may also use multiple conversions so help manage dynamic range issues. A normally exposed RAW file may have part of the sky for example clipped. The clipping can be recovered by the digital exposure adjustment as the expense of the foreground. The two RAW conversions can then be combined as layers in Photoshop using a gradient or other mask to provide a similar effect to using a gradient ND filter with slide film.

Back to TOC

Capture Sharpening

Previously I have used my standard capture sharpening, similar to Canon's recommendation of USM R0.3, T0, 300%. The actual sharpening is a layer edge masked USM R0.7, T0, 300% blended at 65%, using a shareware action from The Light is Right (TLR), Capture Enhanced Single Edge USM [part of TLR Sharpening Toolkit].

However recently I have changed to using a method based on the book Real World Sharpening with Adobe Photoshop CS2 - Bruce Fraser. I have some actions available for download based on this book.

The standard sharpening is for an 8Mpixel source pre-sharpen plus medium frequency image content sharpening. This provides somewhat more sharpening than is required to overcome the camera's sampling aperture (dominated by the anti-alias filter) as is a good process for the majority of 20D images. For some cases slightly different image content sharpening of noise reduction may be used.

This makes this relatively complex sharpening procedure easy to do. The action includes the layer blend setting to protect the shadows and highlight detail, the edge mask protects areas of flat tone from increased noise.

The final sharpening intensity can be adjusted using the blend intensity.

For high noise pictures or pictures, or where the detail is confined to a small area (such as some macro work), I may omit or weaken the capture sharpening and compensate creative sharpening on particular areas.

Back to TOC

Creative and Spot

Spotting if need is performed before capture sharpening. Then creative sharpening and blur if needed. If performed all layers will be preserved and saved as a psd file.

I mostly use the clone tool for spotting.

For creative sharpening and blur I use layers prepared with the TLR Sharpening Toolkit. This prepares the effect on a layer copy of the image with a black mask ready for painting in the effect in the regions required.

Focus Stacking

Previously I used an automatic programme CombineZ5 for stacking. Now I do this manually using masked layers.

Although more lengthy this provides more direct control and gives rise to far less artifacts. Like many things it is more difficult to write down than to do.

CombineZ5

I used the CombineZ5 Focus Stacking Software, mostly with macro shots. This can read TIFF and JPG files and save 8 bit TIFF and JPG files. The software does not preserve ICC, EXIF data or other File Info fields so these need to be restored. The following methodology is used:

a) After making RAW adjustments convert and apply sharpening.

b) Convert to full resolution 8 bit TIFF in the AdobeRGB colour space.

c) Combine in CombineZ5 using the Stack Macro and save as TIFF. The naming convention is to use the original file name with the fff sequence numbers appended fff-fff for an inclusive range or fff fff fff for multiple non-adjacent ranges.

d) (Optional) Make corrections to the depth map if needed (see below).

e) Cleanup any imperfections and crop out any edge effects at the image perimeter, assign the AdobeRGB colour space and insert any file info notes and other relevant metadata. Save as a master in 8 bit TIFF or PSD format.

Correcting the Depth Map

Sometimes the depth map dithers about unnecessarily or creates duplicates of parts of the picture that have moved slightly. These issues can be corrected by the following procedure:

i) In CombineZ save the Depth Map [File->Export Depth Map] This is saved as a colour coded .BMP file.

ii) Open in you preferred image editor. Select the colour you want to paint (in CS2 use the Colour Sample Tool (Eyedropper icon) to set the foreground colour by clicking on the appropriate part of the map.) Then select a soft edge Brush Tool in Normal Mode with 100% Opacity and 100% Flow and paint in as required. Save the File.

iii) Return to CombineZ and load the Depth Map [File->Import Depth Map], a menu will ask you to select what channel colour to use, Green seems to work. Your stacked image will immediately be updated.

Manual Layer Stacking In Photoshop

1. In this process copy-paste the individual frames into a Photoshop document as separate layers.

2. Identify which parts of each layer you want to use and align these using the move tool in a pair-wise fashion from the background layer up. To aid alignment set the upper layers blend mode to difference; return the blend mode to normal once finished.

3. Occasionally you may find a layer has rotated a little relative to the others. Use the measure tool on an easily identified line in the image to establish the rotation amount. Then return to the original separate layer and rotate the canvas before copy-pasting this in as a replacement layer. Return to 2 above.

4. If on rare occasions two separate areas of a layer are not aligned simultaneously due unavoidable changes of scale because of the focus plane shift there are two approaches.

a) Return to the original file and rescale the canvas rather similar to 3 above

b) As scale errors are normally very small just duplicate the layer and align one for one area and the other for the other area. In practice this is faster and easier but does result in a larger file.

5. Now you have all the layers aligned, create a layer mask for each except the background layer. I fill the layer mask with black (none of layer) or white (all of the layer) depending on the situation and paint the mask with the inverse colour using a brush 0% hardness, mode normal, opacity 100%, flow 100% to blend the layers together to form the stack.

6. Having completed the stack, spot the layers as required.

7. Now continue with the normal work flow adding luminance noise reduction and specialist sharpening layers as required.

Below shows the Photoshop layer pallet image for a stacked example with the layer masks:

Each layer is shown starting from the bottom background layer frame 11

Then the masked frame 12 (body and proximal wings and proximal legs)

The masked frame 13 rotated 0.4 degrees counter clockwise (ends of proximal wings)

The masked frame 16 rotated 1.2 degrees counter clockwise aligned for the tail

The masked frame 16 rotated 1.2 degrees counter clockwise aligned for the distal wings

The final stack with luminance noise reduction but no sharpening

Back to TOC

HDR Images

CS2 includes facilities for HDR images. To create my HDR images I set all the RAW exposures to be included to a common colour temperature and open in Photoshop with no sharpening. Then choose Automation->Merge to HDR selecting open files from the popup menu. This gives the option of ticking the align images check box.

Note that using grossly over exposed frames can be detrimental as glare and even lens flare may then be introduced into the 32bit floating point HDR image.

It should be noted than the merge to HDR sets the colour space to the Photoshop default whatever the input picture colour space is. Additionally the resolution is reset to 72 dpi, this can cause unexpected errors during printing if forgotten. This can be changes in Photoshop bit not saved in the Radiant. EXR or PBM file formats. Neither can metadata be stored in that format.

Both these problems can be solved by using the PSD format, however the storage requirements for this are about 4 times that of the the Radient format. Conversion information from HDR to 16 or 8 bit format can be stored in a tonality information file HDT from the conversion window.

My file naming convention is for files called General Tree 001.CR2, General Tree 002.CR2 and General Tree 003.CR2 to name the HDR file General Tree 001-003 HDR.PSD. Converted 16 bit files and the tonality information files might then be named: General Tree 001-003 HDR C01.PSD and General Tree 001-003 HDR C01.HDT another alternative conversion from this file would then be stored as General Tree 001-003 HDR C02.PSD and General Tree 001-003 HDR C02.HDT.

The real issues with HDT are in the choice of conversion to fixed point. As subsequent processing will likely be needed I always convert to 16 bit.

As image magnification can not be changed in the menu I like to work at 100%. There are four conversion method available by choosing Edit->Mode->16 Bits:

a) Exposure and Gamma: This allows you to set the overall dynamic range but not control the tone curve. As a result images are likely to be low in contrast and need masked adjustment of contrast and level on the important subject matter. Clearly masked adjustments like this can result in artifacts at the mask edges and the stretching of the low contrast can cause poor tonal graduation.

b) Highlight Compression: This renders the mid and dark tones wells but as might be expected compresses the highlights. Probably good for previews but nothing else as there is no controls over this process.

c) Equalize Histogram: This seems to effectively compress shadows and highlights resulting in good rendition of mid tones. Again there is no control over this process so it is of little use.

d) Local Adaptation: This is a relatively complex method that has two sets of user controls. Firstly the tone curve can be controlled for black point, white point and a mixture of slopes curves between those points. Also there are two slider controls Radius and Threshold that control the "Local Adaptation" algorithm. This seems to be rather like the USM control and permits the use to control the side and amount of local contrast enhancement. This is probably the most useful method, although it is unclear if it is possible to set the "Local Adaptation" algorithm into a no effect mode but the nearest to that seems to be to set the Threshold to its minimum of 0.1, the Radius control does not seem to do much then. If the local adaptation feature is used care needs to be taken that artifacts are not introduced around sudden transitions of tonality.

Following conversion to 16 bit any subsequent clone stamping is performed and adjustments are made with layers. This is then followed by Capture, Creative and Output sharpening as per the rest of the workflow.

Back to TOC

Output

Web Output

Print Output

Output Resolution

Epson Stylus Photo 1290

Print Workflow

I use two different workflows depending on destination of the image.

Web Output

For web based output the image is re-sampled to the required image size (not more than 640X400 pixels), some high pass mask sharpening is performed, then the image is converted to the sRGB colour space, then converted to 8-bit and saved as a 10 quality JPG.

The sharpening after resizing is based on that used in TLR Professional Sharpening Toolkit for web images (highpass mask radius 0.3 pixels).

This process is often automated. If no creative sharpening or spotting is needed the capture sharpening is prefixed to this process so the automation can be applied from Adobe Bridge directly on the RAW files permitting easy batch conversion to web sized JPGs.

Technically for images down sampled by a factor of 4 (16th the number of symbols) typical of the size used from the 20D, the capture sharpening phase makes almost no difference and could be omitted. But at a down sample factor of 3 or less the capture sharpening is noticeable.

Considering that a factor of 4 down sampling reduces the Nyquist rate to under 20lp/mm from the original almost 78lp/mm and referring to the general characteristics of an anti-alias filter this is reasonable as a quartz AA filter will have a quite respectable MTF of about 0.78 at 20lp/mm on a 20D for example.

I use a modified JAlbum skin for presentation.

Back to Output TOC

Print Output

The printed output workflow depends on the input image size, output image size and the natural input resolution for the printer being used.

Output Resolution

The natural printer input resolution is the highest ppi that can be printed without the printer driver RIP (Raster Image Processor) having to resample the image, this can cause undesirable artefacts.

The website www.inkjetart.com has a test to determine the usable printer resolutions.

This involved two downloadable PDF files that contain test patterns at various ppi. These should be printed at photo quality and examined for Morie patterns caused by resizing in the RIP.

Back to Output TOC

Epson Stylus Photo 1290

This printer is now a little old but produces acceptable quality for general use up to A3+ size.

Most commonly I am using the Epson Premium Glossy Photo Paper with the printer set for the following conditions:

Colour Mode

Photo Quality 1440 dpi

High Speed

No Colour Adjustment

The results for the www.inkjetart.com test are summarised in the table below.

ppi	120	180	200	240	260	280	300	320
result	ok	ok	ok	ok	Morie	Morie	Morie	Morie
ppi	340	360	380	400	480	720	960	1440
result	Morie	[1]	Morie	Morie	Morie	[2]	Morie	[3]

[1] Some faint horizontal banding suggesting alias effects in one dimension although the test pattern was not disrupted.

[2] Morie in one dimension

[3] Aliased pattern

These results repeated with the High Speed check box off although it did change the pattern slightly for [1].

This suggests the highest reliable resolution of 240.

Back to Output TOC

Print Workflow

Following RAW conversion, capture sharpening and any creative and spotting work the following steps are used to print. I do not normally save a print ready copy of the file.

a) The image is flattened to speed up processing.

b) The image is then upsized using bicubic smoother or downsized using bicubic sharper.

c) Some additional sharpening may occasionally be applied. I find this is mostly not needed. If it is, I may use the TLR Capture Enhanced Single Edge USM with the same settings as used in the capture process and adjust the blend to give the effect required.

d) I use the appropriate Inkjet ppi sharpening from the TLR Professional Sharpening Toolkit (for 240 ppi glossy: light and dark high pass masks with radius=1 pixel. Matt is 1.1 pixels radius) to compensate for the printer dot gain.

e) I print via “print with preview” using Photoshop to control the conversion to a printer/paper colour space with “Perceptual Intent” and "Black Point Compensation" selected.

Back to Output TOC

Back to TOC

Calibration

I am not at this time using personally calibrated colour spaces for the monitor, printer or camera due to having better things to spend money on. I do use Adobe Gamma to set the monitor Gamma to 2.2 and the monitor is adjusted for a 5000K white point.

Despite the crudeness of this I get quite predictable results with the subjects I shoot.

Back to TOC

Last Updated 11/09/2008

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.

Any trade marks mentioned are the property of their owners.

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