There are many programs for showing tunnel/cavity surfaces and sometimes other data such as profile plots. A non-exhaustive list:

• Caver and Caver Web, described in Petřek et al., BMC Bioinformatics 7:316 (2006) and Stourac et al., Nucleic Acids Res 47:W414 (2019), respectively.

  Example in recent publication: Ext Data Fig 9 (Nature 2022 Oct 13) made with the Pymol Caver plugin server first finds pockets using Voronoi tessellation, reports their volumes and estimated druggability... ...then allows choosing a pocket as starting point for tunnel calculation download zip includes scripts for Pymol and VMD method uses points on a grid, finds tunnels that minimize an expense function based on proximity to atoms relatively fine grid needed to limit errors from grid approximation, but is computationally expensive for large structures (e.g. ribosome), hence MOLE...

  For ChimeraX visualization, see:

  • ChimeraX recipe for CAVER display
  • chimerax-users Apr 2023: size commands to set each radius
  • chimerax-users Jul 2021: python script for setting bfactor as radius

• MOLE and MOLEonline Server, described in Petřek et al., Structure 15:1357 (2007), Sehnal et al., J Cheminformatics 5:39 (2013), and Pravda et al., Nucleic Acids Res 46:W368 (2018).

  Chimera showing 1bl8 pore from MOLEonline

  channels classified into tunnels (cavity → exterior) and pores (exterior → exterior) method uses Voronoi tesselation, finds pathway along Voronoi edges that minimize an expense function (as above) each channel is represented by a set of spheres

  • web server results – screenshot [pore mode example – screenshot]:
    • can show table of properties for all channels, or details for a specific channel: list of residues, profile of some property vs. length (radius, hydropathy, etc.)
    • other cavities (open to exterior) and voids (not open to exterior) from Voronoi tesselation are shown as spiky surfaces
    • hovering the cursor over a tunnel gives its length and bottleneck radius
    • download zip includes scripts for Pymol, VMD, and Chimera
    • the Chimera download is a session file with the protein and its channel(s) and cavities – molechannels-1tqn.py
    • export from Chimera to ChimeraX omits the cavities (surfaceModel not yet supported) – molechannels-1tqn-export.py
    • TomG enhanced ChimeraX to read the “JSON” download and display channels, with slider if more than one – molechannels-1tqn.json, molepore.cxs (hide largest spheres, e.g. hide @@radius>5)
  • Pymol and Chimera plugins are GUIs for running standalone MOLE 2.5 (instructions; haven't tried)
  
  
• MolAxis server described in Yaffe et al., Nucleic Acids Res 36:W210 (2008) and method in Yaffe et al., Protein 73:72 (2008).

  ...details in Chimera image gallery

  like MOLE, uses Voronoi diagram and is much faster than Caver good agreement with HOLE results, paper Figs 2,3 some novel aspects for faster computation: all atoms treated as balls of same size to avoid curved algebraic surfaces uses medial axis of molecule-unoccupied space to convert pore-finding from 3D to a 2D problem better than MOLE at not giving multiple instances of nearly same channel tutorial describes how to view in VMD

  
  
• ...and what about CASTp?? Tian et al. Nucleic Acids Res 46:W363 (2018). It is focused on pocket detection and all kinds of useful measurements (volume, surface area, circumference, number of openings, etc.). Visualization is associated but not a strong suit.
  • There is a CASTp database and calculation server
  • Like MOLE and MolAxis, uses Voronoi tesselation and the related concepts of Delaunay triangulation and alpha shapes (details...)
  • Chimera has a CASTp fetch and nice pocket-browser GUI (e.g. open castp:1bl8) but you can see the limitations (from a visualization standpoint) of simply using atomic surface patches

The next two claim to be higher-resolution:

• HOLE, described in Smart et al., J Mol Graph 14(6):354 (1996) [PDF].

  Oliver's example image

  finds a possible route for a ball squeezing through the channel, using plane-by-plane Monte Carlo to find largest radius that doesn't intersect with atoms better for straighter paths such as ion channels; for more tortuous paths, may produce zigzags and even discontinuous lateral jumps (according to the MolAxis authors) I couldn't run ancient binary (32-bit from 2015) and didn't want to deal with source, although it's freely available from github under the Apache-2.0 license Oliver Clarke's bash script for visualizing HOLE output in Chimera is described in a previous post (Aug 2015) seems popular amongst crystallographers; documentation describes how to view in VMD and Pymol Examples in recent publications: Structural modeling of the hERG potassium channel and associated drug interactions Figs 1 & 3 Structural mechanisms of TRPV2 modulation by endogenous and exogenous ligands Fig 1 Activation mechanism of the mouse cold-sensing TRPM8 channel by cooling agonist and PIP2 Fig 4ab

  
  
• HOLLOW, described in Ho and Gruswitz, BMC Struct Biol 8:49 (2008).
  • fills cavity with dummy atoms and uses their collective surface
  • hrmmm... *was* available at hollow.sourceforge.net, but disappeared just days ago! (Oct 2022)
  • fork by Beckstein lab is available on github

This one is a bit different, in that the output is a map:

• ³V: Voss Volume Voxelator described in Voss and Gerstein, Nucleic Acids Res 38:W555 (2010).

  uses two-probe “rolling-sphere” method: volume excluded by small probe (~ molecular surface) subtracted from volume excluded by large probe (~ convex hull) server generates map files of pocket and channel “void” volumes that can be shown as isosurfaces in Chimera (or ChimeraX or other map visualization software)