TuFuse Pro Documentation

TuFuse Pro Homepage

Contents Introduction

TuFuse Pro is a Windows "Image Fusion" program that can be used to create extended depth of field ("focus blended") images, extended exposure range ("exposure blended") images, and other types of composite images from a series of aligned input images. TuFuse Pro uses a technique known as Image Fusion to combine the "best" (e.g. best exposed and/or best focused and/or most saturated, etc.) regions of multiple images into a single "fused" composite image.

While TuFuse Pro offers a lot of configuration options, if you simply want to produce an exposure or focus blended composite image from a series of images, then the default options should give satisfactory results under many circumstances. Of course, to get the most out of TuFuse Pro, you may need to experiment with the settings described in this document.

Overview

This section presents a brief overview of image fusion. A high level understanding of the concepts here will help make sense of the controls available in TuFuse Pro.

The concept of image fusion is not new, nor is the more specific type of fusion performed by TuFuse Pro: Multi-resolution image fusion. In fact, the approach was described at least 25 years ago. The term "multi-resolution" refers to the fact that images are fused (sometimes called "combined" or "blended") at multiple frequency levels and each of these frequency levels are added together to produce a single, final composite image. Although it may not seem obvious or intuitive, an image can be decomposed into "frequency bands", similar to the way in which an equalizer on a music system allows a listener to adjust the volume of certain frequencies in the music.

In multiresolution image processing, the image is decomposed into multiple "levels", where each level contains lower frequency details than the previous level. Each level is itself an image that contains only the information in one particular frequency band. Each level is half the size (i.e. half the width and height) of the previous level, and the images produced by this decomposition would create an imaginary "pyramid" if they were stacked on top of each other. The lowest/bottom level of the pyramid is the largest level, and contains the highest frequency details. The highest/top level of the pyramid is the smallest, and contains the lowest frequency details.

TuFuse Pro decomposes each input image into a "pyramid" of levels, resulting in as many pyramids as there are input images. Each pyramid has the same number of levels (the number depends on the size of the output image, and is usually somewhere between 8 to 12 levels for images produced by most commonly available digital cameras). Next, TuFuse Pro constructs a composite pyramid by using the "best" pixels (or a weighted average of "high quality" pixels) from the input pyramids. At each level in the composite pyramid, TuFuse Pro combines pixels from the corresponding level of each input pyramid. Lastly, the composite levels are collapsed into a final, fused image.

How does TuFuse Pro decide how to combine pixels from the input levels? What consitutes the "best" pixel or a "high quality" pixel? The answer depends on what type of fusion is being performed. However, in general, the contribution of a pixel to the final image depends on its "quality". For example, when performing focus blending, "high quality" pixels are those that are in well focused regions. Similarly, when performing exposure blending, pixels with "high quality" are those that have medium intensities (i.e. not dark black or bright white). For each pixel in each level of each input image TuFuse Pro calculates three "quality" measures based on how (1) well exposed, (2) well focused, and (3) saturated that pixel is. These three measures are combined into a single measure of quality based on the weight assigned to each quality measure. For example, when performing focus blending, the weight assigned to the focus quality measure is 1 and the weights assigned to the other two quality measures are 0. In this case, the only quality measure that has any impact on the final result is the focus quality. When creating the levels in the composite pyramid, pixels with low quality contribute little (or not at all if only the "best" pixels are used), while pixels with high quality contribute more.

For more detail on image fusion, and links to more reading, consult the TuFuse web page. TuFuse is a free command line program upon which TuFuse Pro is based. In terms of image processing, TuFuse and TuFuse Pro work very similarly, although TuFuse Pro adds a number of improvements, features, options and capabilities not available in TuFuse.

Main Screen

The main screen of TuFuse Pro allows you to load images, configure program settings and start the image fusion process.

File operations. To load images into TuFuse Pro, you can either (1) use the File -> Add Images... menu item, (2) use the "Add..." button, or (3) drag-and-drop files from Windows Explorer onto the Input Files list view. Selected files can be removed by using the "Remove" button and moved up or down in the list by pressing the "Move Down" or "Move Up" buttons. TuFuse Pro can read TIFF and JPEG files. If DCRaw is installed in the same directory as TuFuse Pro, then it can also read any RAW files supported by DCRaw (see here for more details). Note that DCRaw is not required in order for TuFuse Pro to operate, but it is required for TuFuse Pro to be able to work with RAW images. DCRaw is not distributed with TuFuse Pro, but is freely available on the internet.

Batch processing. By default, all images that are loaded into TuFuse Pro are fused into a single output image. However, TuFuse Pro can work on "batches" of images, fusing images in one "batch" into one output image, images in a second batch into a second output image, and so on. Images can be assigned a batch ID (by default all images are assigned a batch ID of "1"), and each batch of images is processed separately. To assign one or more images a unique batch ID, select the desired images in the list view, and press the "Assign Batch ID" button. TuFuse Pro will assign the selected images the next unused batch ID.

Output files. Use these options to choose whether to specify the output file name(s) manually or have TuFuse Pro create a name autiomatically. If you specify an output name, and request TuFuse Pro to process multiple batches, TuFuse Pro will append a numeric identifier to the name you choose (e.g. name_0001.tif, name_0002.tif) to ensure that the output files from each batch do not overwrite each other. TuFuse Pro can output files in either TIFF (8 or 16 bit) or JPEG format, configurable via the file browser that appears when the "Browse..." button is pressed. If TuFuse Pro chooses an output name, it will use TIFF format. The filename that it creates is based on the first and last image names in a batch, and the file is saved in the same directory as the first image in the batch.

To show the configuration screen, press the "Configure Settings" button.

To start the image fusion process, press the "Fuse Images" button. TuFuse Pro will start fusing all batches one at a time until all images have been processed. Once the processing starts, the caption on the "Fuse Images" button changes to "Cancel Processing". If the button is pushed while TuFuse Pro is processing, then processing is stopped. Note, however, that the batch that is currently being processed will run until completion, although any remaining batches will not be processed. While processing is happening, the status bar at the bottom of this screen will indicate what batch is being processed, and also show the status of each batch as it is being processed.

Configuration Screen (General Options)

The configuration screen is accessed by clicking the "Configure Settings..." button on the main screen. The "General Options/Preview" tab allows for the adjustment of parameters that guide the fusion process, and shows a preview of the fused image. At the right of the screen are several boxes with a blue caption bar. Depending on the size of your screen, these boxes can be expanded or collapsed as necessary by clicking on the blue caption bar.

Fusion Mode Presets

Autobracket

TuFuse Pro can fuse each input image with a lighter and/or darker version of itself. When the "darken" or "brighten" sliders are set to non-zero values, then darker and/or brighter versions of each input image are created and fused with the original image(s). This can be useful to boost shadows and/or constrain highlights in a single image. For example, fusing a single image with a brighter version of itself will produce a final image with brightened shadow regions.

Fusion Pyramid Levels Adjustment

This slider adjusts the number of pyramid levels used during the fusion process. TuFuse Pro automatically selects a default number of levels, and this slider can be used to adjust the number of levels relative to that default. The "default" number of levels that can be used on a series of input images is automatically determined by TuFuse Pro, and depends on the dimensions (width, height) of the output image. For this reason, the default number of levels for a preview image is usually less than for a full size image (larger images can use more levels).

Weights/Iterations

These options are only available in custom preset mode. You can choose the number of iterations to use by selecting one or two as the iteration count. If two iterations are selected, then the first iteration will fuse images with similar exposures (see also the Miscellaneous options screen to configure how TuFuse Pro detects similar exposures) in groups, and the second iteration willl fuse the images resulting from the first iteration. If only one iteration is selected, then all images are fused in a single pass.

The weights govern how much TuFuse Pro assesses the "quality" of each pixel in the input images. Quality measures are based on well-exposedness ("exposure"), contrast (used as a proxy for focus) and saturation. For example, a weight of 1 for exposure and 0 for contrast and saturation means that TuFuse Pro only evaluates exposure when determining the "quality" of each pixel in each image. Similarly, values or 0.5 for saturation and 0.5 for exposure means that TuFuse Pro assigns equal importance to both measures when determing the "quality" of each pixel. Each pixel is assigned a quality by TuFuse Pro using one or more of these three measures.

The "combine" boxes (more precisely, "Laplacian combination mode") determine how TuFuse Pro combines pixels in the input images into a final composite. If "average" is selected, then all pixels in all input images contribute to the final result, where the contribution amount of any pixel in an input image is determined by its quality (See discussion of weights above). Pixels with a higher quality measure contribute more heavily to the final result. If "Maximum" is selected then only the "best" pixel (as measured by its quality...see discussion of weights above) is used.

Note that the determination of quality, and combination of pixels is performed at multiple levels (i.e. frequency bands) in an image.

Exposure Weight Curve

The exposure weight curve is used by TuFuse Pro to determine the "well exposedness" of each pixel in the input images, one of the three quality measures that can be used by TuFuse Pro to guide the fusion process (see discussion of quality measures in the weights panel above). The horizontal (x) axis shows the intensity/brightness of a pixel, and the vertical (y) axis shows the corresponding quality assigned to a pixel. The default curve shape resembles a "bell curve" which means the very dark (left edge of the curve) and very bright (right edge of the curve) pixels are assigned a low exposure quality, while pixels with a medium intensity (center of the curve) are assigned a higher quality.

The sliders in this panel allow for the adjustment of the shape of the curve, changing how TuFuse Pro assesses the exposure "quality" of each pixel. For example, moving the brightness slider to the right shifts the curve to the right, and means that TuFuse Pro will assign higher quality values to brighter pixels than the default curve shape. There is no right or wrong shape for the curve...your artistic goals should govern your choice of curve shape. In many cases, the default curve shape gives a well balanced blending of images with a wide range of exposures.

Slider descriptions:

Focus Blend Options

These options are only relevant when performing focus blending, and are only available using the auto or focus blend preset mode.

Highlight/Shadow preservation. Focus blending a large number of images into a single composite can sometimes have the effect of increasing the contrast of the final image...shadows become darker, and highlights become brighter. (This is because TuFuse Pro uses contrast as a proxy for focus when assessing the quality of a pixel, and the fused image is a result of combining the most "contrasty" pixels from each input image.) In some cases the highlights or shadows may even become "clipped", meaning that they become featureless white or black regions with no detail. This control allows TuFuse Pro to reduce or eliminate "clipping" if it occurs as a result of focus blending. Higher values eliminate clipping more completely, but can reduce overall contrast. A value of zero performs no highlight/shadow preservation at all. Note that this control only has an effect when clipping is detected by TuFuse Pro. In the event that focus blending does not cause any clipping, then this control has no effect on the final image. Also, if the input images are clipped themselves (i.e. contain regions of featureless white or black), then this control will not resolve this problem.

Perform exposure blending on upper levels. If selected TuFuse Pro will perform exposure blending (rather than focus blending) on the upper levels of the fusion pyramid. The advantages of this approach are(1) it helps mitigate the amount of increased contrast/brightening that can result if all levels of the pyramid are focus blended and (2) allows for better focus blending of images with differing exposures. Because the upper levels of the pyramid contain only very low frequency detail, there should be no apparent loss of shaprness/focus in the final image when using this option.

If performing noise reduction, do so... These options determine when noise reduction should be performed (see the noise reduction panel for more information on noise reduction). If noise reduction is performed during focus blending, then TuFuse Pro can analyze all of the input images in the batch and gather additional useful information to guide the noise reduction process. If noise reduction is performed after the focus blending is complete, then it only analyzes the resulting composite image to perform noise reduction. In general, better results are usually obtained by performing noise reduction during the focus blending step. Note that this option does not turn noise reduction on or off, but rather adjusts at what stage of TuFuse Pro's processing should the noise reduction be performed. If noise reduction is not enabled in the noise reduction panel, then these options will have no impact on the final image.

Levels/Histogram

This panel shows a histogram of the currently displayed preview image. The red, green and blue channels of the image can be displayed individually or together using the checkboxes above the histogram.

Beneath the histogram is a horizontal line with three adjustment controls ("sliders"). These sliders can be used to adjust the image's levels (i.e. the tonal range) by manipulating the intensities of the image's shadows, midtones, and highlights. Levels adjustment is performed on the composite image that TuFuse Pro produces by fusing the input images. In other words, TuFuse Pro performs fusion first, and then levels adjustment on the resulting image.

The black slider modifies the image's black point. Any pixels that are darker than the value set by the black point are set to black in the adjusted image. Similarly, any pixels that are brighter than the white point (set by the white slider) are set to pure white in the adjusted image. Any pixels with values between the white and black point are redistributed across the range between the white and black points. Moving either the white or black sliders away from their default positions at the ends of the sliders increases the contrast in the resulting, adjusted image.

The midpoint (grey) slider adjusts the distribution of the pixels between the white and black points. Moving the slider to the left produces a brighter image, and moving the slider to the right produces a darker image. The degree to which this brightening or darkening of the image affects the dark and light pixels in the image can be adjusted using the Highlight/Shadow bias slider. If the Highlight/Shadow bias slider is moved towards the right, then any adjustments made to the midpoint (grey) levels slider will have more impact on the shadow regions of the image (i.e. shadows are brightened or darkened more than highlights as the midpoint (grey) slider is moved. Conversely, if the Highlight/Shadow bias slider is moved towards the left, then any adjustments made to the midpoint (grey) levels slider will have more impact on bright/highlight regions of the image.

Noise Reduction

This panel controls how/if noise reduction should be performed on the image. If enabled, noise reduction can be performed on the "luminance" and/or "color" channels of the image (Luminance is a measure of the brightness of an image).

Detection Threshold. This control determines how TuFuse Pro detects noise. Higher values will result in TuFuse Pro identifying more pixels as "noise", which are subject to noise reduction.

Reduction Amount. This control determines how much noise reduction should be performed on those pixels that TuFused Pro identifies as "noise". Higher values result in more aggressive noise reduction, while lower values will leave some remaining noise.

Levels. This control determines the number of pyramid levels on which noise reduction should be performed. Lower values (fewer levels) result in noise reduction only being performed on high frequency noise, while higher values (more levels) result in noise reduction being performed on high and lower frequency levels. It is often the case that color noise in current digital cameras has a lower frequency than luminance noise, and so the default number of levels is different for color and luminance noise.

By default, noise reduction is performed on the composite image that is created during the fusion process. However, if TuFuse Pro is performing focus blending, it can be configured to perform noise reduction during the fusion process (See the focus blending options). This noise reduction uses a different approach, and uses the additional information available from examining all input images in the batch to guide the noise reduction process.

Preview

The preview panel can be used to generate a preview fused image for the lowest numbered batch of images using the currently selected settings. TuFuse Pro can either generate a small preview or a full size preview. Small previews generate much more quickly than full size previews and can be useful to guage the results of exposure blending changes. Full size previews take more time to generate, but are more useful in examining the results of focus blending and/or noise reduction changes.

TuFuse Pro maintains a "history" of each preview it generates, and the "<" and ">" buttons on the left and right of the preview button allow you to navigate backwards and forwards through this history, quickly comapring the results of changes to settings.

At the left of the preview panel are a series of checkboxes that can be used to include or exclude images from the preview.

If checked, the show console checkbox will cause TuFuse Pro to show the console output from TuFuse (the command line program that does the actual image processing) as processing is occuring.

Configuration Screen (Advanced Options)

Output File Options

The following options configure settings relating to the output file(s) created by TuFuse Pro. To specify either TIFF or JPEG output, use the file browser dialog invoked by the "browse..." button on the main screen.

Tiff Bit Depth. Specify bit depth for output TIFF files.

Tiff Compression. Specify compression (lossless) used for output TIFF files.

JPEG Quality. Specify quality (inverse of compression amount) for output JPEG files. High quality = low compression, and low quality = high compression.

Miscellaneous Options

Image Gamma. The gamma parameter is used by TuFuse Pro when calculating the relative exposure of images. The default value of 2.2 is appropriate for most images.

Exposure Difference Threshold. This parameter specifies the maximum exposure difference (in stops) that TuFuse Pro will allow and still consider input images similarly enough exposed so that they are focus blended during the first iteration when using the Auto blend preset.

Preferred Processing Color Space. TuFuse Pro can be configured to process images in RGB color space or YCbCr color space. In most cases, this will not make much difference to the appearance of the final output, although some images may show slight color differences depending on which color space is used. You may wish to experiment to decide which result you prefer. Note that some operations (such as noise reduction during focus blending) require that a specific color space be used (e.g. YCbCr), so TuFuse Pro may override the selection made here if it needs to do so to perform a necessary operation.

Keep focus blended images after first iteration. if using the Auto blend mode, and TuFuse Pro performs focus blending, this box will cause TuFuse Pro to retain the focus blended images (in the system temporary directory) when complete. If not checked, these focus blended images are considered to be transient, itermediate images and deleted upon completion of the final image.

Blend across -180/+180 degree boundary. If checked, TuFuse Pro will perform fusion between the left and right edges. This is useful for 360 degree images, where the left and right edges are intended to be joined together.

DCRaw Options

These options all modify the way in DCRaw is invoked by TuFuse Pro. For complete information on what these options do, please consult the DCRaw documentation.

In order for TuFuse Pro to invoke DCRaw correctly, the following conditions must be met:
  1. DCRaw must be installed in the same directory as TuFuse Pro. TuFuse Pro will not locate any other instances of DCraw that may be installed on your machine.
  2. The filename for DCRaw must be named "dcraw.exe". TuFuse Pro will not be able to locate or use DCRaw if it has another filename (e.g. dcrawms.exe).
  3. The version of DCRaw that you have installed works correctly on your operating system and with your particular types of raw files. Note that Windows Vista, in particular, does not operate correctly with many commonly available versions of DCRaw.
Menu Options

Following are a list of menu choices.

Bugs and Problems

As hard as software developers try, most programs contain some bugs. If you find a bug in TuFuse Pro please feel free to contact the author (see TuFuse Pro's page for contact details).

When reporting a problem, please give as much information as possible. A good bug report is one that allows the developer to reproduce the problem using the steps that you provide. A step-by-step recipe that allows the developer to replicate the problem is likely to result in a quick diagnosis and resolution. Reports with vague wording such as "I saw an error message" or "it doesn't work" make the diagnosis and resolution difficult or impossible.

If reporting a bug, please explain what happened (how many images, what format, what dimensions, what were you trying to do, what buttons did you push, etc.), and provide the TuFuse Pro configuration information (this is the information shown when using the Help -> Configuration Information menu item). Please do not send your images unless the author asks for them. Most importantly, please provide a step-by-step recipe that can be followed to elicit the problematic behavior...if the problematic behavior can be replicated, it can usually be solved. If not, a solution is often difficult or impossible.

More information is always better, and is more likely to lead to a solution...Thanks!

Downloads

Purchasing a License and Registering TuFuse Pro

TuFuse Pro is free to download and use. However, until it has been registered, the images that it generates contain a number of small watermarks. Use the registration screen (the Help | Purchase/Register... menu option) to purchase and/or enter a registration code. The license code for TuFuse Pro can be purchased online (using the "Register online..." button), and that license code should be entered into the registration screen's input fields.

All license codes are purchased online using the Kagi purchase processing service.

Links and More Reading