> [...] the proposed method [...] is tailored to content that puts importance to well defined lines/edges while tolerates a sacrifice of the finer textures.
and
> [...] a big weakness of our algorithm [...] is texture detail, however since upscaling art was not our main goal, our results are acceptable.
That sounds like a multiobjective optimization problem. If this multiobjective optimization problem was solved (permitting the nature or structure of the multiobjective optimization problem, of course), then the algorithm would be improved, don't you agree?
Did the authors of this algorithm not have the capability to formulate or recognize the multiobjective optimization problem?
Or if they did have the formulation capabilities, but that they did not have the capability to solve the multiobjective optimization problem? Why if so? Too difficult? Not enough time? Limited by a resource? No intention to have done so, excepting that they said that a specific trade-off was acceptable?
You're welcome to share your speculation or opinion, Hacker News reader.
I believe they recognise the problem is a multiobjective optimization problem (hence the formulation of their sentence) but their algorithm is not parametrizable : it is a single point on the pareto front and you would need other algorithms to explore the rest of the front.
and
> [...] a big weakness of our algorithm [...] is texture detail, however since upscaling art was not our main goal, our results are acceptable.
That sounds like a multiobjective optimization problem. If this multiobjective optimization problem was solved (permitting the nature or structure of the multiobjective optimization problem, of course), then the algorithm would be improved, don't you agree?
Did the authors of this algorithm not have the capability to formulate or recognize the multiobjective optimization problem?
Or if they did have the formulation capabilities, but that they did not have the capability to solve the multiobjective optimization problem? Why if so? Too difficult? Not enough time? Limited by a resource? No intention to have done so, excepting that they said that a specific trade-off was acceptable?
You're welcome to share your speculation or opinion, Hacker News reader.
I'm curious to know your thoughts, is all.