This article contains a tentative list of future versions and features.

Argument scope

• Fix bugs in argument scope handling.

Multiple-pass registration

• Revise alignment algorithm to allow multiple passes for less sensitivity to noise and better estimation of tonal certainty.

3D scenes and depth recovery

• Revise 3D scene reconstruction algorithm to perform acceptably on a wider variety of scenes.

PSF windowing and infinite-support PSFs for Irani-Peleg rendering

• Add syntax for PSF windowing and new PSF types with infinite support, including the gaussian kernel, suggested by HJ Hornbeck.

Variable lighting

• Add a facility for handling variation in lighting between input frames.

Interchange formats

• Add support for a commonly-used interchange format, such as autopano pto, suggested by Peter Miller.

Raw file handling

• Use Dave Coffin's dcraw program to extract information from raw files, including pixel data, as well as black level and other meta-data, as suggested by Gregory Maxwell.

• Add a generic 2D transformation class, supporting partial transformations.
  • 'partial transformation', as used here, means that it is expected that some points will fail to map.
• Add a partial 2D transformation class with constructor taking as arguments a pair of 3D projective transformations and a 3D model.

• Add new file-scoped options
  • Irani-Peleg
  • black level
  • alignment type
  • focus details

• Modify existing Irani-Peleg code to use generic (partial) 2D transformations.
  • in combination with the above, this should allow the algorithm to operate on 3D models.

• Add facilities for handling depth-of-field and focus bracketing in input image sets, as suggested by Viktors Rotanovs and Gregory Maxwell. This change will likely involve:
  • using the 3D model for depth data, as suggested by Gregory Maxwell
  • using user-provided (file-scoped) focus details to determine depth of field
  • using variable PSF in the Irani-Peleg algorithm, as suggested by Gregory Maxwell
  • developing an approach to handle the incremental (non-Irani-Peleg) case.

• Allow specification of output image characteristics, including exposure (e.g., gamma) and channel offsets within a pixel.

• Add support for specifying output devices, such as LCD monitors, to allow shorthand customization of exposure (e.g., gamma) and other characteristics, such as channel offset in the case of LCD screens for sub-pixel anti-aliasing.

• Add support for specifying weight and certainty upper bounds for rendering, either as an absolute constraint or relative to average weights, so that the overall contributions to an output pixel (or channel) are constrained by weight.

• In combination with weight and certainty upper bounds, time locality could be used to determine which frames are used to generate an output pixel. As a special case of this, video stream processing could favor frames having the smallest time difference from the target frame.

• Refine rendering and alignment code for better performance.

Synchronized frame set syntax

• Add syntax to identify synchronized frame sets.

• Enhance the command-line interface in preparation for i18n. (?)
  • e.g., syntax for identifying synchronized frame sets
• Internationalize the command-line interface and help text to accommodate translations.

• Update the Technical Manual for information regarding 3D operation.

• Generate correctly indented book-level tables of contents when rendering the complete set of manuals.
• Add a tutorial, and credit Metin Toyran for suggesting a step-by-step guide.
• Remedy meaningless page numbers for book title pages in the set ToC.
• Complete migration of existing documentation from the localhost sub-directory, removing superfluous license terms in the footers.
• Centralize licensing and authorship information (e.g., there should be only one COPYING file and only one AUTHORS file).
• Include license and copyright terms for the doc tree Makefile as copyright notices in the AUTHORS and/or COPYING files (or equivalent).
• Complete the documentation.
• Revise the user manual for new features and syntax.
  • Document new scoping operators {}, [], and [<>].
  • Document other changes in UI, including regularization of syntax.
    • in particular, commas can no longer appear in single-argument options when using assignment-delimited syntax; use space-delimited syntax instead. This behavior is inconsistent with 0.8.4.

• Add facilities for manipulating images storing color as integer-valued data, possibly as library routines.

• Move image-processing routines into a library that can be used by other programs.

• Add bindings for other programming languages.

• Investigate options for distributing a 'complete' package including full revision history, pre-built documentation (if the ordinary distribution does not include this), and, possibly, example image sets. Direct download, bittorrent and CD are the most prominent candidates for distribution.
  • In the case of CD distribution, information should probably be added to the manuals regarding how to update CD-provided program source.

• Investigate options for creating and distributing a combined source and binary Windows package containing sufficient software to ensure support for common image types (e.g., by statically linking with libraries for opening JPEG and PNG files). BitTorrent and CD are the most obvious candidates for distribution.
  • This package could build on the package described above, as a 'for Windows' variant or some such.
  • This package could make use of an application such as InstallJammer or Nullsoft Scriptable Install System for installation.