The authors present an cost-effective method to reproduce 3D rockfall trajectories from and CAVR. It has the advantage that uses the 3D terrain model as the spatial reference for the falling block instead of the videocameras and considers a proper offset of the center of mass in the reconstruction of the trajectories. Nice sutdy, clear presentation and meaningful results.

Only a few comments:

1) In my opinion, the sentence "only dissipative impacts are observed with CAVR..." should be removed from the summary. The authors to the SLF experiments. The sentence may be misleading because although it refers to the SLF experiment, the reader may interpret it to be generalizable to other experiments.

2) Lines 34-35 For ease of reading, it is convenient to explain here the difference between the rebound model parameters and the apparent coefficient of restitution.

3) Section 3.3. As the authors mention, it is necessary to evaluate the geometry of the rock block to adequately compensate the impact positions to the center of mass to obtain the correct trajectories. However, the reconstruction the 3D model of the blocks is not fully described. It seems that the model is obtained from the frames but there may be hidden parts of the blocks that can affect the reconstruction of the shape and volume, as well as the measured dynamic parameters. It is convenient to explain in more detail the procedure followed to reconstruct the volume of the blocks and also provide an estimation of the errors.

4) There is an issuet that has not been treated by the authors but that appears in other experiments. It is the presence of dust during the impact. Has dust been generated? If so, how have the authors resolved this circumstance? Is it the algorithm that determines the point of impact and the kinematic parameters?

5) Please, check the references, some references are incomplete or the source can not be easily identified (e.g Berger, 2011; Domaas, 1995; Garcia, 2019; Girardeau-Mountaut, 2006; Sanchez 2020; ...)

Thanks for this nice contribution and the approch to significantly improve video analyses of rockfall trajectories. I have more or less only small comments.

General comments:

• You state that your analysis procedure has some problems with longer lasting block-ground contacts. Would it help to assume that such longer contact periods are not reflecting a single contact but two single short impacts right one after another? Often, these “double impacts” are responsible for the cases where the COR > 1.
• Would it be helpful for your purpose to rely on existing data regarding camera/lense distortions such as e.g.
  
  https://argus.web.unc.edu/camera-calibration-database/
  
  https://lensfun.github.io/
• If you have another camera with different viewing angle how much does the (probably) missing synchronization of the different camera influence the analysis?
• Your manuscript almost everywhere uses qualitative expressions only such as “low, high, good, well, higher, lower, longer, slightly, close, ….”. I would love it to have also some quantitative values as well. Especially in the discussion section. This would lift the quality of the article to a higher level.

Specific comments:

L33: Maybe, you can add references to these two software tools?

L36: Maybe, you can emphasize more clearly that the concept of COR in rockfall trajectory modelling probably is a model only. The rocks themself almost don’t jump (just let a rock drop and observe almost no rebound). So, the COR approach is not reflecting the correct physics behind but compensates the natural edges and corners of blocks and underground that force the jumps.

L56: The posterior analysis of traces in the field might deliver good restaurations of trajectories. Of course a lot of work but often necessary for example in case of heavy damages. Suggestion of additional publication regarding physical trajectory parabola analyses

• Gerber, W. (2019). Naturgefahr Steinschlag–Erfahrungen und Erkenntnisse. Eidg. Forschungsanstalt für Wald, Schnee und Landschaft WSL, Birmensdorf. WSL Berichte, 74, 149.
  
  https://www.dora.lib4ri.ch/wsl/islandora/object/wsl%3A19475
  
  (in German only, relevant pages 17-22, 33-54)

L62ff: The advantage of the blocks of Caviezel et al is their regular shape. This eases the determination of the centre of gravity.

Caption Fig. 1: “minimize” --> “minimizes”

L134: Till which travel speed would you recommend that the air resistance is still neglectable?

Equations 10-11: Maybe, you have to explain the difference of these two CORs in details. Most readers usually knot COR_t and COR_n only.

Figure 3: Can you somehow visualize in the figure that the relation between pixels and nature changes also for the different sections of a recorded image due do lense distortions and the geometry?

Caption Figure 3: “helps distinguish” --> “helps to distinguish”

Equation (21): The letter “a” has already been used for the acceleration. Maybe, you want to change it?

L603: “helps save” --> “helps to save”?

Figure 9: The two diagrams are difficult to read. I don’t know whether I understood them right, but if so then I would recommend:

• Remove “segment above”
• Integrate Fig.9a,b,c, etc
• Remove CAVR and SLF 2020

L488 and others: “Fig.” --> “Figure” if reference to a Figure is used within the text. Abbreviate “Fig.”. only if used within brackets.

L535: remove comma after “Caviezel”

Suggestion of additional references regarding video analyses in the field of rockfalls but using definitively “antique” techniques:

• Glover, J., Denk, M., Bourrier, F., Volkwein, A., & Gerber, W. (2012, April). Measuring the kinetic energy dissipation effects of rock fall attenuating systems with video analysis. In 12th Congress INTERPRAEVENT (Vol. 1, pp. 151-160).
  
  http://www.interpraevent.at/palm-cms/upload_files/Publikationen/Tagungsbeitraege/2012_1_151.pdf
• Glover, J. M. H. (2015). Rock-shape and its role in rockfall dynamics (Doctoral dissertation, Durham University).
  
  http://etheses.dur.ac.uk/10968/
• Volkwein, A., Brügger, L., Gees, F., Gerber, W., Krummenacher, B., Kummer, P., ... & Sutter, T. (2018). Repetitive rockfall trajectory testing. Geosciences, 8(3), 88.
  
  https://www.mdpi.com/2076-3263/8/3/88/htm

References:

• The first two references contain author’s institutions in brackets. I think these can be removed?
• Volkwein et al (2011) is mentioned a couple of times in the text but is missing in the references list. Please, add.
• References Garcia (2019) and Sanchez+Caviezel (2020) miss information on where these references can be found. Please, enhance.
• The usage of “last accessed” currently is inconsistent with the different links provided.