Posts by Ian C

Is there a line of code that can be added to the TMC file so that it rejects (or simply doesn't produce) any masked tiles? I thought I saw it somewhere on this forum, but maybe I was imagining it? I know you can do it with Geoconvert Helper by unticking a box - but I was just wondering if you could do it "manually"?

Hi Andrew

Thanks for your reply. It's great to get another perspective on these things. Regarding question 3, I guess the principle is similar to increasing the render scale factor in the VR settings. The improvement can only be marginal, but it is there. (Not that increasing RSF works for me, but that's another matter!) Like you, I've never noticed any step change to a lower resolution when gaining altitude so you're probably right on that one. The screen door effect is indeed a distraction - mostly, for me, when looking at the distant skyline.

So far I've created 23 GB of tiles converting an area about the size of an English county! So I can see myself having to invest in extra drives if I carry on at this rate.

Cheers

Ian

These are rather long and detailed questions, but if anybody can answer even one of them I'd be very grateful!

1. FSET resolution 3 is 4 m/Pix and resolution 4 is 4'm/Pix. What is the difference?

2. What is the resolution of the best Virtual Earth images? Since they can be captured at 0.25 m/Pix, I assume there are many places where it is at least as good as that?

3. In another post I argued that a level 15 tile is about 800 x 800 metres. Given that tiles of all levels are 2048 x 2048 pixels it follows that each level 15 pixel is about 40 cm x 40 cm. So level 15 is needed to fully take advantage of FSET resolution 0 (0.5 m), as shown in @Rodeo's chart in another thread. That being the case, my question is about the justification for capturing FSET images at a resolution of -1 (0.25 m). My feeling is that, although the final resolution will only be 0.4 m, the quality of information that goes into each pixel will still be better if the FSET resolution is 0.25 m. Does that make sense? [The downside, of course, is that it tales 4x as long to capture the info from FSET.]

4. When I look down (using FSET -1, level 15) I can see sheep in the fields. If the resolution is 40 cm, that means each sheep is an elongated blob of about 1 x 3 pixels - 0.4 x 1.2 metres being my estimated size for a sheep! So far, so good. What puzzles me, however, is that I have the impression that I can see detail smaller than a sheep - i.e. the detail looks smaller than the width of a sheep. Examples are: the white lines on roads (about 15 cm wide?), power cables between pylons (about 5 cm wide?), etc. If there is a 15 cm white line on blackish asphalt I would expect a greyish pixel to be produced with a width of 40 cm. My brain then interprets this as (a) white and (b) narrower than a sheep because it knows that, in reality, a road marking is white and narrower than a sheep. Is this what's happening - or is there something I don't understand about images?

5. Still on the subject of being tricked by the brain ..... Using FSET -1 and level 15 low buildings and trees look 3D to me above about 100 metres. I can't decide whether this effect is enhanced by shadows, or whether it would be the case even without shadows in the FSET images. (Unfortunately this doesn't work for taller structures above about 2 storeys, which look flattened from any altitude.) So it seems to me that I only need cultivation when I'm below 100 metres or above a town where there are taller buildings (or if I need night lighting effects etc - or if the native FSET quality of the image isn't good enough to produce the 3D effect.) Is this how other people see it, or is it just me?

6. The Oculus Rift has a resolution of 1080 x 1200 and a field of view of 110 degrees. This means that the angular resolution is about 0.1 degrees. (Compare this with the angular resolution of the unaided eye, which is about 0.02 degrees or 0.0003 radians. So the Rift would need to improve its resolution by a factor of 5 in order to be equivalent to normal vision.) 0.1 degrees corresponds to a 40 cm (level 15) pixel viewed from a height of about 230 metres (800 feet). So, if you're using the Rift, it's worth having level 15 tiles at any altitude below 230 metres. Also it means that you wouldn't really want to see any lower level tiles (14, 13, etc) below about 230 metres. Is there a danger that these lower levels will kick in when you don't want them (i.e. below 230 metres) thus lowering the optimum resolution? (Given that I only fly low and slow I'm trying to optimise my scenery for low altitude.)

I understand that the scenery tiles (all levels) are 2048 x 2048 pixels. At mid latitudes a level 10 tile is about 16 x 16 miles, level 11 is 8 x 8 miles ..... down to level 15, which is about 0.5 x 0.5 miles. (0.5 miles = 800 metres.) If you divide the size of the tile by 2048 you get the pixel size on the ground. I guess the pixel size/resolution of the highest level used needs to be smaller than the FSET resolution in order to fully take advantage of the latter? Also the overall resolution can't be any better than the FSET resolution, no matter how high the level of the tiles. That's the way I see it anyway - but I'm a relative novice, so could be wrong.

Quote from Spit40

Everybody Cultivate !!

Ian C , I urge you. Wait not a moment longer. Make do with your interim geoconversions and start cultivating. There is nothing like it. I just went for a night/dusk flight through the Lake District in the Peruvian plane. I was living the dream. I see military jets fly in the Lakes fairly often and the realism in FS2 with cultivation is amazing. The streetlights and twinkles from clusters of houses is so convincing especially at night. It totally changes the scenery to have these buildings responding to ambient light or generating their own in a realistic way. Use my scenproc file and you won't need to inspect the OS Data, just do the reference system conversion in QGIS.

My geoconversion is still crude. I missed a few white patches at sea and I should really do it again at 0 or -1 (I processed at +1) but it makes much less difference once you cultivate.

OK thanks Phil - you've convinced me! I think I can probably move on once I'm happy with the way I deal with coast - which hopefully will be fairly soon

Spit40 Yes It would be interesting to have a place to discuss specifically UK related matters. On Goodle Earth some of the most scenic parts of the country (eg the Lake District, Scotland, etc) have rather washed out images. I haven't checked Virtual Earth/Bing, but I imagine it's the same there. And yes - I did see your post on the use of OS data and cultivation. That's something I'll probably turn my attention to once I've fully mastered the basic geoconvert thing. (I thought I understood it, but I still find that areas of scenery sometimes don't show up - and I can't discover any rational explanation!)

Spit40 Reading the thread again I realise that what I'm suggesting isn't radically different from what you were suggesting, except that I was proposing using the FSET .bmp sea colour near the coast to paint in the white areas - rather than using the default FS2 sea colour. The default FS2 sea colour could also be determined (if desired) by taking a snapshot in Aerofly and maybe one could try to transition between the two sea colours several miles from the coast. Anyway I'm in the process of trying it now and will report back later.

The idea of using the tidal data sounds great, especially if you're a bit of purist about the real images as I am.

I'm currently working on scenery in SE England and so far - being a novice - I've been avoiding the sea altogether. However, as the OP rightly says, if you're in the UK at some point you have to take the plunge (in my case into the freezing cold waters of the English Channel).

Admittedly I haven't actually tried this yet, so I may be talking nonsense - but I thought I'd just put it out there as an alternative way of dealing with the coast. (I will be trying it shortly, once I've grabbed the images I need.) What I'm thinking is that I could leave the FSET sea image alone on the .bmp file as far out to sea as it goes and, if there's an ugly white patch, simply copy the colour of the FSET sea near the coast into into the offending patch. I might even be able to copy the texture as well - I haven't played enough with Gimp to be sure of that. In fact why not fill the whole sea right across to France with the same colour/texture? I'm thinking if I grabbed a big area in the middle of the Channel (or wherever) from FSET it might be all white - or black - I'm not sure, I haven't tried it. But anyway, I could fill it with the FSET sea colour/texture that I had near the coast and then geoconvert it. All this could be done at low res.

Like the OP I reject all masked tiles, so this would deal with the problem of masked tiles on land being indistinguishable from those produced by the alpha process applied the coastal .bmp files. (That was my understanding of the problem. Correct me if I'm wrong.) It also kicks the problem out to sea, so to speak. Any kind of abrupt transition would be less objectionable out at sea than near the coast where it's potentially spoiling the natural coastal scenery. Maybe I should have waited until I'd actually tried this before talking about it, but I just wanted to get my thoughts down on "paper" before I forget them - and maybe others can see potential snags before I waste my time.

Rodeo

OK thanks - that's confirmed what I thought. I didn't read your conversation question. Where is that?

Ian

Edit: OK I've found it now and replied. I'm not familiar with all the features of the forum yet!

Yet another question. The more I do this, the more questions keep coming up!

I've now decided to start with a level 10 tile as a base area and build the other levels on top of that up to level 15. My higher level tiles will be generated from FSET images with a resolution of -1. I'm wondering if there is any point in using FSET images of a lower resolution for the lower level tiles (10, 11 and12)? I understand that the lower level tiles need to be of lower resolution to function properly, but surely they will be of a lower resolution anyway since they are a similar file size to the higher level ones but cover a bigger area? The fact that they are generated from a higher resolution original doesn't affect that - or does it? I can understand that if you were only creating levels 10, 11, 12 it would save time to set FSET to a lower resolution - but since I'm doing the whole area at -1 anyway I might as well use the higher res for the lower levels.

Rodeo

Many thanks - that's very useful. It looks like the origin is at lat 90 S and long 180 E/W, not 0,0 as I was thinking.

Ian

Quick supplementary question to add to my list .....

There are 4 level 1 tiles covering the whole earth. Do they all meet in the middle at lat 0, long 0? If that is the case it has just occurred to me that it would be relatively simple to calculate the position of any level of tile at any point on the earth's surface.

(Just to be clear, I am assuming that each level 1 tile has dimensions of 90 deg in latitude and 180 deg in longitude. So a level 2 tile is 45 deg in latitude and 90 deg in longitude, and so on. Hopefully that is correct too?)

Hi Rodeo

Many thanks for your very clear answers, as always. I'm now in the process of trying out my own idea and I will certainly let you know how I get on. It's good to know that I can always add the lower levels later if it's not satisfactory.

Cheers Ian

I think I've read most of the tutorials etc on geoconversion, but there are still one or two things I'd like to get clear in my head. I'd be very grateful if anybody can find time to answer any of these!

1. The size of a level 9 tile appears to be roughly 1 deg x 1 deg. (The size in km obviously varies with latitude.) Is it exactly 1x1 deg, and do its boundaries coincide exactly with whole degrees of latitude and longitude? The reason I ask is that one way of avoiding masking problems is to snap to a level 9 tile in GeoConvertHelper. However, unless you know exactly where the level 9 tile is, you would have to gather images from FSET in an areas of roughly 2x2 deg to be sure of capturing one whole level 9 tile.

2. Do the boundaries of the higher level tiles coincide exactly with boundaries of lower level ones? (I assume they do.) e.g. Are there 4 level 10 tiles precisely within one level 9 tile, etc etc?

3. I assume that the higher level tiles are not visible beyond a certain height/distance, so that your computer does not have to process information that is too small for your eye/monitor/VR headset to resolve. Is that correct? Does anybody know roughly what these distances are for different levels? The reason I ask is that I shall mostly be gliding at no more than 1000 m above the ground. Could I get away with just levels 13, 14 and 15 for example? (I could determine this by experiment, but I was just wondering if anybody already knows the answer.)

4. Do the black (masked) parts of lower level tiles obscure higher level tiles that exist in the black region? Is this why masking causes such problems? However I assume you would still see the higher level tiles if you flew low enough?

5. I always set FSET to a resolution of -1 (0.25 m). This means that it would take 24 hours at the very least to gather images to cover a whole level 9 tile. An area of 0.2 x 0.2 deg (roughly 20 x 20 km) seems more reasonable time-wise. My thinking is that one way forward for me would be to set GeoCovertHelper to give me no masked images (by unticking the box). This would mean that I would only get maybe level 12 tiles and higher within a 20x20 km box, since tiles at all levels lower than 12 would be too big. They would fall partially outside the area and would therefore be masked - and hence rejected. If I wanted to extend my scenery area I could then geoconvert another neighbouring 20x20 km square with a 50% overlap with the first 20x20 km square. I could then dump all the files into the same folder and reject the duplicates. This way I ought to get complete and seamless coverage at level 12 and higher without any masking problems. Does that make sense?

OK thanks Rodeo - I've done it again and it works this time! Not sure what I was doing wrong before

Although I have successfully created one scenery area using Geoconvert I have not so far made use of GeoConvertHelper. I've downloaded it, unzipped the folder and changed the extension from .ex to .exe but I can't seem to be able to run it. I've double-clicked it and used the "run" command, but to no avail. It appears to be a notepad file - is that right? I'm not very knowledgeable about computing so I'm sure I've missed something very simple. Any ideas?

Quote from Rodeo

Hi Ian,

This is fine. You have overwritten the area with the same tiles, so they only exist once in aerofly.

This is an explanation of tile masks, and what happens during conversion:

Same TTC file names

The best strategy is to avoid masks by snapping to tile borders. This can be selected in the GeoconvertHelper tool by snap to Level 9 borders.

But Level 9 tiles have an extension of rough 78 km, so in smaller areas you will get masks unavoidably.

Rodeo

Display More

Many thanks for your reply Rodeo. I'll study it and try it out later today.

Since I wrote the above post I've read a bit more in the forum, and I think I now have some understanding of what the masked tiles are. (They're the ones at the edge of the area, so they're not full tiles?) If this is the case, would it be better to get rid of them altogether? If you did that I imagine you would need more overlap between areas. What would be a reasonable overlap for 2 adjacent 18 x 12 km areas containing levels 9, 11, 12, 13 & 14?

I've got to a similar stage - i.e. successfully made one 10' x 10' area of scenery in southern England (roughly 18 km x 12 km at this latitude). I'm using a resolution of -1 in FSET and levels 9, 11, 12, 13 & 14 for the whole area (as opposed to different areas for different levels). FSET seems to take a couple of hours to produce the .bmp images and geonconvert takes about 30 mins to do its job. I had previously created a small test area a few km away and, like the OP, I copied the new larger area into the same folder as the smaller area, choosing to replace files with the same name. When I selected "Location" in FS2 I could see the new larger area but not the old smaller one.

Reading this thread it seems that I might be able to solve the problem by putting the different areas in different folders. I've noticed that many of the geoconverted files have another companion file with the same name, except for "_mask" at the end. What is the purpose of these mask files? Do I need them? Could I dump all tiles in the same folder if I got rid of the mask files? These mask files don't seem to exist in the default FS2 or DLC scenery.

Many thanks BTW to all who have made the this possible - whoever wrote the Wiki, whoever produced the video I watched & the help tools I used.

Hi overloaded. Yes I have used the pause key to do that, but I think it's the randomness and quietness of balloon flight that appeals to me - never being quite sure where you're going to end up. Also the fact that you can land it in odd places - a bit like a helicopter. Of all the current aircraft the ASG 29 glider is the best for viewing scenery in my opinion, but I imagine it would be great to lean over the basket of a balloon and look down with no intervening glass.