Need Feedback on Water Effects

[=Hollywood=]

Hey Guys,

I've got a bit of a challenging situation with water mapping and I think it could affect other projects as well...

I hope it's okay to post the ideas here as well as I'm really looking for feedback on the best way to solve water issues with FSX...

In developing wide area photoscenery for FSX we are coming up against a problem that really wasn't a concern in the days of FS2004/2.

The issue at hand is water mapping data. Years ago FS used to just use a solid 0,0,0 RGB value in an image to define where water was located. No problem, paste in black water and the default FS2004/2 water shows through.

In FSX it's a different story because FSX uses alpha channels in the source imagery where black = the water movement effects and white = land. Thereby overlaying water effects but allowing imagery underneath to show through, which would be great if all water variables in the real world stayed constant.

In other words, imagery that was taken of the ground in 2007 will see very big differences in water levels at dams, streams and rivers etc than water mapping data that may have been obtained in say 2009. With tidal bulge, evaporation, droughts, floods etc these water variables create problems for water mapping on a massive scale.

So I wanted to bring this dilemma to the table with the FSX community and solicit feedback on the most effective solution to this particular water mapping problem issue for massive scale rendering. Below is a screenshot of our alpha stage work and the arrows point to some of the types of anomalies likely to be experienced with just the water effect placed over the imagery...

Now, here's the thing. We have a few choices:

• [li]We can publish with these anomalies and risk the ire of the FS perfectionists who expect more than what is technologically possible today[/li][li]We can paint the terrain a uniform deep blue color which will lose the subtle variances contained within the aerial image based water but will still give an effect similar to the water in X-Plane and will make the water more of a unique feature[/li][li]We can do both 1 & 2 above and allow users to choose which they prefer[/li][li]Long Term we may be able to punch out the water from the terrain and allow FSX default water class polygons to show through that we'd develop specifically to match our photoscenery, but this option would take an extraordinarily long time to implement, yet would be the final option we could give users in future[/li]

My personal thoughts are that Option 1 is great for the start and get the imagery rolling and out the door. Option 2 can follow shortly after. Option 3 would likely be available to start rolling out 1 year from now.

We do have plans to eventually run image analysis tools over these datasets in future but they are very expensive right now and require massive amounts of computer power which would only really be available through cloud computing.

So what do you guys think?

[edit:]

I'll start the discussion by mentioning a couple more things...

The USGS data that is being used by others is terribly inaccurate and doesn't include data such as these type of evaporation ponds...

The other remaining photosceneries that have high accuracy water are limited to small areas where they've been able to manually paint in the water masks to be super accurate, which is something not feasible on USA wide photoscenery...

So with those things in mind, that this water dataset contains a huge amount of data compared to anything else currently available worldwide, what are your thoughts about water coverage, accuracy and anomalies... Given that the main alternative out these doesn't even come close to this level of water mapping detail...

Basically what would you find as an acceptable approach to this problem?

Note: another option would be to have a shader expert tweak the FSX shaders to make the blue tinted water wave effects more like clear water...

 

[=Hollywood=]

The blue alternative would look more X-Plane-ish like you see from this shot from an X-Plane addon product...

 

[arno]

Hi Dean,

I would indeed say your last option, making them transparent and have water polygon laying under it, would probably the best to go with in the end. If you have well enough vector data of the water it should not be too hard to use it for both the terrain and the alpha channel I guess.

If the quality of the photos is very good (so not too much colour differences between shots, etc). It might be interesting to keep the water as shown in the image since that will show more details in the water.

My last approach would be to paint all of the water blue. Of course with the watermask you can still get some wave effects, etc on top of this, but it will look less nice. And when we tried this approach we had trouble with lines showing up due to the compression of the images.

 

[GHD]

If you have hydropolys, you could use a grey blend mask to reveal the underlying colour and a water mask where you need wave effects.

The attached is a quick and dirty blend I did for part of Guam.

 

[pina47]

You could also feather the borders of the water bodys, most ponds and stuff like that pick up idrt on it's borders, and loose it's relfectiveness, not to mention they won't have a water stream for sure.

 

[=Hollywood=]

If you have hydropolys, you could use a grey blend mask to reveal the underlying colour and a water mask where you need wave effects.

The attached is a quick and dirty blend I did for part of Guam.

Well that stuff's the basics and we already have those in place but the main issue is in the accuracy of the data and how the inaccuracies are too apparent imho...

However you did just give me an idea on something that could be effective in the long run... If I were to do option 1 and option 2 but for option 2 create that layer with a % of transparency that might just be the ticket...

It's taking so much thinking out of the box to come up with an effective solution...

I still like the thought of tweaking the shaders though to get rid of that blue slushie tint...

 

[=Hollywood=]

Wow, well render the entire discussion above as a completely moot point...

I have found *THE* solution that leaves all these other solutions looking like an 8086 computer...

My world has just been rocked!!!

 

[=Hollywood=]

Saw something interesting yesterday with regards to the blue tint in FSX...

I'm starting to revise my thoughts, for FSX development I think we need to throw a white DC3 into various types of water and see what the tint of the water is...