Question about User Sketching

[DanielSHKao]

Dear Authors,

Thanks for your inspiring work! I am reading through your paper, aiming to build my own project based on the work. However, I found that the paper does not include many details about user sketching. Specifically, I expect that the reconstruction performance will vary on different sketching quality and views. Could you please elaborate more on how to select the views for sketching? Are there any empirical or technical rules/tips to decide which views should be selected for sketching? For example, I expect that sketching views with more occlusion and far from each other (in terms of camera poses) could result in less view-inconsistency and higher qualitative results (as the in-painting model 'fills' more details?) intuitively. Is this idea correct?

Thanks.