Speed adjustment considerations

[Just Crazy Jim]

All very sensible. The data regarding the PopTop Locos isn't terribly surprising. I've played with them for years and never had any hard data, but what your testing demonstrates is in line with my experience. And I have never been content with the performance of the locos in RT3, the models they chose were all good for their time, but that is seldom reflected in any comprehensible manner in RT3. Some, like the Duke, were even kept in service through two or more transitions. There were even a few that started as freight (the EMD GPs for instance), then switched to non-express passenger service when the freight got too heavy (If you have ever ridden an Amtrack train outside the DC-to-NYC belt, you know what I mean), then short line service when the next model got demoted, then finally relegated to yard dog service as a switcher.

Some of the mechanics that operate in RT3 are striking anachronisms. I cannot speak for how things were in the "golden age", but these days, in the USA, the notion of a naming a locomotive to increase profits is a bit like clapping gold earrings on a pig to increase the price to the consumer for the bacon made from it. But, then again, people flock to any number of gold-plated hog troughs for "superior dining experiences" (koff*koff*Appleby's - sneeze*sneeze*Red Lobster). Also, the mechanic of a "station" is quite "a century ago", most freight is loaded at and delivered to a location that is not a station per se. And, yes, I know that RT3 is not a rail simulation, etc. And, yes, I understand that sacrifices had to be made to make a playable game (I am still playing it). But, be that as it may, there is the conceit that RT3 represents real railroading in some pedestrian manner and the places where it drifts from reality, such as representing iconic locomotives poorly, is part and parcel of the general drift from realism.

I want the game to be something more, as, I believe, we all do. I am all for making the trains run more profitably. I am all for making the iconic locomotives do in-game as well as they did in the real world. I am all for making the locomotives being field assigned in-game as they were in the real world. But after trying to tamper with the game's economics, I am becoming increasingly hesitant to suggest that more realism is what is needed. So, allow me to play the Devil's advocate a moment. Please, keep in mind that none of this is meant as a criticism or meant to dissuade you from hammering at the game files.

Off-Topic:

Take, for instance, the matter of rail miles as calculated by the ledger. If we agree that 1 track segment equals 1 statute mile, then we must also agree that 1 height map pixel equals approx 1.37 statute miles. That, then, makes us need to face the elephant in the room, and admit that:
1\ The game's height exaggeration for terrain is astonishingly out of whack and that no locomotive yet made could scale the peaks on most maps without catastrophic failure, and,
2\ The average firebox of a steam locomotive is approximately 1 mile wide, thus inferring that the quantity of fuel cost needed to move such a locomotive from the World Trade Center Station in New York City to Grove Street Station in Jersey City, New Jersey - a distance of approximately 1.7 miles - would exceed the GNP of a fair number of small nations. These two things alone dismiss all doubt that realism was most assuredly not what they designed the operation of locomotives around in developing RT3.

Now compound all that by the fact that the game's "mile" is a flexible, somewhat misleading, unit of measure that may or may not, on any given map, represent the same unit of measure in terms of the distance between the World Trade Center Station in New York City to Grove Street Station in Jersey City, New Jersey. The unsettling reality of that is that a finely calibrated locomotive that is set up for large maps with long distance runs may spend 50% of the year loading and unloading on another map. Even though a locomotive that is not in motion does not have fuel cost, it still ages, steadily decreasing fuel efficiency, and still has a maintenance cost. True, though it may be, that the flaw is not on the locomotive's side of the equation, but rather bears on the map-maker, the player will most likely not have a wide spectrum of options to choose from in terms of locomotives based on the year of the map. The knee-jerk fanboy reaction is "make more locomotives for the game!" (and I likely will), but how then do we go about calibrating new locos to fill the niche for maps that are peppered full of short-distance runs without making the new locomotives useless clutter in the purchase window on larger maps?

I have had in the past, because I neglected to add a maintenance shed to a line, had locomotives that ran dry on oil and remain dry on oil for six or more years until I happened to catch my oversight by merit of the "out of oil" icon displaying in the train list. After reading the information on the forum about oil and its relationship to reliability and chance of breakdown, I wondered how that was even possible. It has led me to think that "chance of breakdown" approaches being a meaningless term when Random Number Generation is in play. RNG is not like drawing cards from a deck of cards with a known and limited number of cards, so that each card removed increases the certainty of a specific card appearing as cards are removed. Rather, RNG operates more like a card being drawn at random from a freshly opened deck, every time. So, the expression "chance of breakdown" lacks the terminal certainty that it implicitly expresses. Conversely, that also means that the reliability rating approaches the same problem. I believe this to be true because the way the game expresses a reluctance to generate a new random kernel over multiple runs of a map in a given period of time (i.e., the same random tycoon assigned to the player, strikingly similar building arrangements during the seeding of the map). So, if what is observationally true extends to the hidden parts of the game that rely on RNG, it is entirely possible that the RNG gets "stuck" in a certain range until some intervening circumstance (e.g. rebooting the system) and thus possible that even the most reliable locomotives operate as badly as the most unreliable, with very negative impact on the profitability of that "near perfect" locomotive. I have had "bad runs of luck" in this direction over the past 14 years, and my solution has been to quit the map, reboot the system and start over.

How this chance of breakdown affects the performance of all locomotives on a map is best illustrated when one does breakdown and brings all other locomotives on that track to a halt. This would be a very significant problem if RNG does indeed get "stuck". More so, if the goals of the map are such that the player needs to haul X loads of Y cargo within a given period of time and the plan goes down the pipes over one stalled loco (I'm looking at you Russia map). In such an instance, fuel economy and grade climbing and acceleration are somewhat less crucial than getting the goods delivered without interruption of service. And it drives home the fact that we have very little control on how profitably we can make any given locomotive operate if we cannot with certainty control the chance of breakdown of all locomotives more perfectly or at the very least more fully understand how the RNG works and how the game calculates and effects the chance of breakdown.

With many of the game's functions locked away from view in the EXE, without calling them liars, I am beginning to wonder if any of the information we have been given by PopTop is reliable. The Big Boy is a prime example of the PopTop Locomotive Chart being downright misleading. It brings into question almost every part of the economic model. At start of play on any given map, the price of coal begins at a set price. If a locomotive's primary fuel is coal, then it stands to reason that the fuel cost should then be linked to the market price of coal within the game's economy. Only it isn't. The same is true of diesel locomotives; their fuel costs are the same regardless of the market price of diesel. And I have no idea what sort of pixie dust is involved with electrics, because electric plants seem to be nothing more than cargo sinks. Thus, the very conditions that caused steam to become obsolescent in the real world may not evolve on a game map, yet steam power is edged out regardless of that. But, then that begs the question, what manner of steam locomotive would be working rails in a world of 80 or more ton freight wagons? Or, even, how do we represent "alternative fuel" types (e.g. liquid natural gas, hydrogen, nuclear) in the game without recourse to game events and custom language files that are dependent on the map-makers to use?

From my own very limited experience tampering with game files, my concern would be for Johnny "Ima download all this" Newbie compulsively downloading and installing everything without fully understanding how new locomotives (and possibly new buildings) are going to impact the playability of PopTop scenarios and campaign maps. Many of which are already terribly skewed.

Now, with that out of the way, a heartfelt bravo! Carry on, this game needs more of your genius in it. And please, dear Gumboots, make the Duke and Stirling recolours available for download.

[Gumboots]

That's all fine and dandy mate, and I agree with some of it as far as it goes, but is something like this:

1/ People want more choofers. Can't have too many. Train nuts play RT3, and train nuts like trains. Wotcha gonna do?

2/ It aint never gonna be perfect. It doesn't have to be. It just has to be entertaining.

3/ The map size thing isn't much of an issue in practice. Distance between stations can affect choice of locos (acceleration matters more for short hops, top speed more for long runs) but doesn't seem to bork things big time. People have played all sizes and types of map and it all works, or near enough.

4/ I agree that reliability is often more important than nominal speed. I tend to play to minimise breakdowns. I've found breakdowns to generally work the way they should, and yes I do understand your point about probability (I've explained it others before).

5/ Fuel cost is completely unrelated to map economy, and that's that. It's one of RT3's simplifications and there's no point worrying about it, so I don't worry. I thoroughly recommend this brilliant strategy.

6/ Steam isn't going to be edged out by the time I'm done with things.

what manner of steam locomotive would be working rails in a world of 80 or more ton freight wagons?

7/ Big grunty ones, of course.

8/ Yes, alternative fuels are a nuisance, and will inevitably require anyone who wants to play them to do a bit of brain engagement. The same goes for various ideas about a 21st century economy, instead of RT3's basically 1950's economy.

9/ Newbies who want things simple can keep things simple. If they don't have enough sense to do that, they will learn the hard way. It's pointless trying to make things idiot-proof, because someone will always build a better idiot.

Now, with that out of the way, a heartfelt bravo! Carry on, this game needs more of your genius in it. And please, dear Gumboots, make the Duke and Stirling recolours available for download.

Will finish the best ones at some point. At the moment they're good from a distance but need detailing. But I'm not feeling possessive about them. I'm quite happy to throw a PSD at anyone who wants to get into them. Or to finish one in particular if there's a high demand for it.

[RulerofRails]

4/ I agree that reliability is often more important than nominal speed. I tend to play to minimise breakdowns. I've found breakdowns to generally work the way they should, and yes I do understand your point about probability (I've explained it others before).

Well, in RTII breakdowns last 6 months, the 1 month variety don't seem so bad after encountering those. But, I agree with your point nominal speed doesn't mean everything. Other factors can be enough to offset a higher speed. That was the reason behind this thread in the first place. I think I made a little progress such as calculating that the average loco is stopped for service/load/unload for 1/3 of the year or more. But I met with the issue of differing usage. Some sort of approximations are needed to find a set of "guideline-principles". I got to a point, but then I started to question if others used their locos as hard as I do.

Jim, I'm not aware of any plan to fake loco stats in the current roster so that all gaps are filled. There will always be gaps for new locos. What I hope is that we can come up with some documentation which together with the spreadsheets is simple enough for a loco maker to get an idea of how to make their loco perform the role of their choosing without the "whatever-looks-good" approach used in the past.

My main motivation for a re-balance of the engines is to bring their economic/performance ratio closer together. Maybe things will be a little more realistic, and progress in this direction is good, but it's a game after all so improving game-play is higher priority. I'm quite tired of playing with only the most efficient (according to RT3, not real-life) locos. At the moment I tend to play a whole scenario with one or maybe two engine types, because game economics dictate that.

As for making the locos more profitable, let me explain a little. In RT3, the economy (prices on the map and therefore price differentials are hard-coded throughout the whole time period 1829-2000+), more cargoes are introduced as time progresses, but the actual economics involved in a train hauling a load is the same for all the time periods. For example, Alcohol has the same revenue potential (if it has the same demand/supply environment) in 1830 as in 1986. On the other hand, loco fuel costs (esp. for steamers) go up with time, one can see some big jumps which are the default era changes, 1850, 1900, 1950, but otherwise the increase we see is due to increasing loco weights on the steamers more than anything. Take a look at this.

I'm all for getting for partially leveling out the hump in the steam engine line. I did make some rough proposals somewhere, but am interested to see what Gumboot's instincts are to avoid steam getting "edged out."

Put simply, my feeling is that fuel costs in the early period are actually a bit low, sometimes it's just too easy to buy many extra locos, so I would increase those a bit. Then definitely the steamers in the 1900-1950 on average need to be cheaper to run (not hard to notice as Gumboots did, that the Pacific has hideous running costs). My idea is to base "guide-lines" to cluster locos around what is the current usable efficient loco. For example the H10 is usable in my opinion, but much higher costs I think we are just pushing players toward industries. In the average game, trains always compete with industries for your investment. It's not too hard to get 20% ROI from industries at least for the first 5-10 years of the typical game. I manage to push some excellent short-term returns from rails (closely tied to engine purchase price), but there's scant evidence of others doing this. So not a good thing to base decisions on.

[Gumboots]

Re the default roster, I just had a look at that from the steam-only perspective. It's a bit odd. Here's the (pre-CtC) chart with my best attempt at loco classification. There are a lot of locos which were express in real life, and a shortage of usable grunters. Then there's the fact that a lot of the express locos are a bit nobbled too, at least from the late 19th century onwards.

The CtC expansion helps a bit. The QJ fills a hole for a usable grunter for the mid-to-late 20th century, but for express in that period unless you only have Class 01 until the Red Devil appears. The 242 A1 and Class U1 are more oddities than anything else, as they aren't available for long and aren't actually that usable. The LNER A1 isn't bad if you can stand looking at it, but again isn't available for all that long.

I'm all for getting for partially leveling out the hump in the steam engine line. I did make some rough proposals somewhere, but am interested to see what Gumboot's instincts are to avoid steam getting "edged out."

Basically, make more of them to fill gaps, and/or extend the availability timeframes and tweak stats on some of them, and give them performance that makes them usable, without going mental. What else can you do?

Put simply, my feeling is that fuel costs in the early period are actually a bit low, sometimes it's just too easy to buy many extra locos, so I would increase those a bit. Then definitely the steamers in the 1900-1950 on average need to be cheaper to run (not hard to notice as Gumboots did, that the Pacific has hideous running costs). My idea is to base "guide-lines" to cluster locos around what is the current usable efficient loco. For example the H10 is usable in my opinion, but much higher costs I think we are just pushing players toward industries. In the average game, trains always compete with industries for your investment. It's not too hard to get 20% ROI from industries at least for the first 5-10 years of the typical game. I manage to push some excellent short-term returns from rails (closely tied to engine purchase price), but there's scant evidence of others doing this. So not a good thing to base decisions on.

Basically agree with all of that, but would be interested in what you think regarding tweaking stats to allow your style of use for locos. Pros and cons.

[Gumboots]

By the way, if anyone wants it I added some tools to the speed vs grades spreadsheet. I'm calling this out of RC and into Final.

Short version is it now has better formatting, some information on actual mileages achieved on various runs by various locos (for fuel bill estimates), and tools to give an indication of useful lifespan for a given breakdown chance and consist weight.

[RulerofRails]

Basically, make more of them to fill gaps, and/or extend the availability timeframes and tweak stats on some of them, and give them performance that makes them usable, without going mental. What else can you do?

Yes, for sure. I was wondering in terms of how their Yearly Average Cost related to the defaults and therefore how this relates to the existing Diesel and Electric costs.

My style is to try to get the maximum from what I have, it wont matter what happens with any set of values, I'll still attempt that. I wasn't planning to do anything to advantage my play-style. Something that really helps with high returns is getting hold of express traffic when I first connect to a city and taking care with the passenger/mail overviews to get an excellent ROI from that (basically load up a train with Any Express heading to the destination that will give most revenue, can be filled further with a switch to any cargo after the game is un-paused for a minute). Expansion needs to be skewed to the closest cities (hopefully large) for this. When considering ROI from rails, engine purchase price is part of the short-term cycle. Think of it like this, I made this connection for say $800k, $200k for a station and then $100k x 2 for trains. The ROI for the first year-ish is based on how much profit my trains can make on their first. If that's $500k that would be a 42% ROI. But if engines cost $200k each, that would drop to 36%. If I build maintenance facilities that adds an extra $130k, dropping this further to 32.7%. This becomes a cycle to finance the next expansion. Somewhere in here note must be made that there may be other opportunities (good price differentials have developed as time passed) for new engines on the network, if I can make more than the engine cost on the first run, it's almost a certain buy. Three such expansions will provide financing for an entirely new one the following year. The higher the ROI, the faster the cycle and therefore expansion across the map. Building Hotels is a great help to maintaining a decent ROI for the next part of the life-cycle for that line. Gives that boost to ensure it maintains at 10%+ while my focus is on new expansions.

I'm not great at explaining the process (I typically fine-tune for each situation), but the things that matter for locos are low purchase price and decent express rating. With the new system, I might have to use two dedicated engines which will be good, change up the strategy. You might see where I got the idea for a Mixed class with cheap purchase price, does both passengers and freight with decent speed but the running costs are not great. For that initial push, I mainly care about short-term returns. I could see myself purposely using engines with more expensive running costs until I had my system stretched far and wide across the map and new expansion opportunities were becoming limited then coming back to replace them, this time taking more care to find a loco designed more specifically for that role. However, reflection on this idea makes me realize that it's not realistic and might actually do more to limit the use of these mixed locos for other players. Which I believe you saw from the outset, so I'm not going propose it again just say that it's a product of the way I play. This ties in with my worries about Mixed class being too strong (I saw your note, and I would agree saying that IMO we currently don't have the loco selection needed to do it properly). But, like I said testing was always needed with the new car weights which will likely influence engine decisions in more than one way I don't anticipate.

Different tangent: (when I talk about "choice" some of that is focused on the short term I mentioned above)
Also, in my play I don't rate performance on grades that highly (I tend to choose the Duke over the Consolidation), unless I have a huge range to climb. Most of the time there is a way to avoid stiff grades through track-laying/routing choices. Speed on a grade is a measure of the maximum speed that will be reached from a standstill after the period of time indicated by the acceleration level. It's much more efficient to have a loco accelerate on the flat and then hit the grade with momentum after which the loco will continue up the grade at the maximum speed indicated. Sometimes I run track a bit longer just for an acceleration area. Maximum speed is sort of a speed limit. As long as a train is capable of 10mph+ on any grade on it's route, I tend to not worry about it the same as I don't worry about breakdowns if reliability is above a certain level. I will tend to double-track (stops for passing on grades are bad, bad, bad even for locos with good hill-climbing) graded sections before flat sections, and also think about the top speed on a potential return downhill trip when figuring. Of course, I'm talking about balance. Higher speed on grades is a definite advantage for that engine, but loco choice is always a decision made by balancing against other considerations. There's also the chance that my behavior has been influenced by the default locos where the hill-loving "freight" locos seem to be in general heavier (worse fuel costs)/higher maintenance/slower/poorer acceleration.

But, yeah, I don't think others think about the short term/long term stuff. Trying to think in a more generalized way for this. But there is the danger to water down too much. . .

Also, how sure are you that the linear increase in breakdown chance past Year 2 has a 1/9 gradient?

Pretty confident. The top part of this is raw testing data. You will notice that some of the unloaded totals don't make sense, the game was displaying them as halved since the trains were in the station (that's because of this bug). But the progression with age, as can be seen by the lower section (showing the multiplier) is solid.

[Gumboots]

Basically, make more of them to fill gaps, and/or extend the availability timeframes and tweak stats on some of them, and give them performance that makes them usable, without going mental. What else can you do?

Yes, for sure. I was wondering in terms of how their Yearly Average Cost related to the defaults and therefore how this relates to the existing Diesel and Electric costs.

In general, I was planning on keeping steam costs a bit more challenging than diesel and (particularly) electric, since we're used to it being like that anyway and it's more realistic (as if that matters in RT3!). But steamers will never match the speed of a Skinkansen, and I wasn't going to try to make them do that. Just get them usable. The Red Devil is usable, even though I don't particularly like it in default form, and even though IRL it was useless on grades, but for all round performance it's in the ballpark. Ditto the Kriegslok and Class QJ.

Also, in my play I don't rate performance on grades that highly (I tend to choose the Duke over the Consolidation), unless I have a huge range to climb.

So I assume that's because of the pax rating for short term expansion, and possibly the better acceleration, since they both have the same running costs with a given consist.

But, yeah, I don't think others think about the short term/long term stuff. Trying to think in a more generalized way for this. But there is the danger to water down too much. . .

I don't think we have to water this down too much. We are aiming at an audience that has exhausted the basics.

Also, how sure are you that the linear increase in breakdown chance past Year 2 has a 1/9 gradient?

Pretty confident. The top part of this is raw testing data. You will notice that some of the unloaded totals don't make sense, the game was displaying them as halved since the trains were in the station (that's because of this bug). But the progression with age, as can be seen by the lower section (showing the multiplier) is solid.

Cool. That's what I used in my sheet anyway. Just thought I'd ask.

[Gumboots]

Put simply, my feeling is that fuel costs in the early period are actually a bit low, sometimes it's just too easy to buy many extra locos, so I would increase those a bit. Then definitely the steamers in the 1900-1950 on average need to be cheaper to run (not hard to notice as Gumboots did, that the Pacific has hideous running costs). My idea is to base "guide-lines" to cluster locos around what is the current usable efficient loco. For example the H10 is usable in my opinion, but much higher costs I think we are just pushing players toward industries.

I just had a great idea over in the other thread. There's a quick and easy way of double-heading Connies. This would make an ideal unit for filling in between the default single Connie and the H10 when default cargo weights double in 1900. As well as being fun. Running costs are still reasonable even if everything is just doubled. It gives the same performance with 20 ton cars as the single has with 10 ton cars.

[Gumboots]

I made myself a chart to see where things are currently at with 1.05 and 1.06 steam locos. I've been ruthless and done what a lot of people do when actually playing: ignore the mutant locos which are just too ugly to use.

Tis a bit gappy, innit. Pretty well covered for express, but hardly anything for late 19th century and early 20th century freight. It's basically the Connie or nothing.

The lack of steamers after the 1960's is easy to fix. I can just extend the availability for any that still have existing examples (which funnily enough would rule out the Red Devil).

And yes, I did leave out the 1.06 Daylight, simply because its stats make no sense compared to the default 1.05 Northern. There was hardly any difference in performance between the GS-2 and GS-4. The Daylight was just a re-paint in real life, and IMO the 1.06 Daylight should just be a skin. Nothing more. As a skin it's currently a mess, so I discounted it.

I've included some of the custom locos that I can stand looking at. Left the rest out. Might have missed one or two. Anyway this gives a good idea of what needs fixing. The G10 is worth having, and I have that sorta fixed somewhere in Blender. Ditto the BR39 if I ever get around to finishing that (really should). I actually like that one. Although really, since it's just another mid-20th century express steamer, you could argue there's no need for it.

The 1.06 Mogul makes no sense at all. Same timeframe as the Connie but less grunt, so what's the point?

Ditto for the G4. There's already the DX Goods, another smallish 0-6-0, which is available for much longer. The G4 seems like a waste of an engine slot. It's not one that I can get enthused about fixing either, while OTOH I already have the DX Goods 95% modelled and almost ready to skin.

Not sure if I'll ever bother with the P-2. The current model and skin bear no relation to a P-2, and the skin is borked, and its stats are mental, so if I was going to go to the trouble of making a freight Mountain for that time slot I'd probably finish off the NSWGR D57 instead. Plus there's already the H10 and the Berkshire in that period anyway.

Ten wheeler? Meh. Vittorio Thunderbox Dunny Shed On Wheels? Meh and ROFL.

Class S? Probably should. V2 and Black 5? Definitely should.

[CeeBee]

Sounds like enough to keep you busy till Xmas I like the chart you made. It about sums it up

[Just Crazy Jim]

The history of European steam is not my strong point.

Where would you place a Webb compound 0-8-0 in there? Before or after the Connie? Better or worse? I wonder, because Baldwin was still manufacturing Connies as late as the 1920s, whereas Webb 0-8-0s didn't start until 1901 and seem to have had a very short run and then went through quite few rebuilds throughout their service that resulted in ever-changing class names (B, E, F, G, G1) - that seems to indicate a problematic design.

And would you say a Midland Railway 115 Class ("Spinner") is the same as the Stirling? It seems on a cursory reading of the specs on Wikipedia that the Spinner might be slightly better, but maybe not enough to bother modelling. They're similar enough in appearance and performance, that I think PopTop may have intended the Stirling to represent both.

[Gumboots]

I wouldn't bother with a Webb compound. They weren't that great. I had already thought of a Midland Spinner, and have one partially modelled. They would be good as a sort of upgraded Stirling for later (heavier) pre-1900 pax consists. This is actually what they were built for. But not essential.

[Just Crazy Jim]

Thanks for the new info.

Off on a new tangent. I've been trying to think of some fanciful way to marry an early European aesthetic to the idea of the Challenger 4-6-6-4 or Big Boy 4-8-8-4. Yes, I know, madness - either a steampunk, hyper-glandular madness with too many 7-foot drivers (why stop at 7?) for 19th century boilers and pistons to move, or a Red Devil/Class A1Nord Superpacific got into the steroids and human growth hormones (again) madness.

I really need to stop following every link I find on locomotive fanboy sites. My grip on reality was already shaky.

[Gumboots]

Right at the moment I'm concerned with getting models of actual locos sorted. If I start going steampunk the results could be deadly.

So I've been playing with skinning again. If I'm going to unbork some of the existing borked ones so I can stand using them, skins are going to be required. This has got me upgrading my (minimal) Photoshop skills, which is good. I'm at the point where I can see how I can get results that I'll really find satisfying. One thing I am sure of is that the default skins were not made at 1024x1024 and definitely not at 512x512. I suspect they were made as 2048x2048 and then scaled down. This would make doing all the detailing on the complex skins a lot easier.

Anyway I've just been chopping the defaults for the moment. Unborking the BR39 will require a good Darth Vader skin (DR black mit red) and a good Prussian railways skin (their original livery). So I made these from the default Class 01 skin, which won't fit the BR 39 properly but are good for practice. The Vader skin is prototypical for a Class 01* though, so can be used on that.

The default skin for the Class 01 has always bugged me too. Those daft white wheels and associated bits look like a candy cane with a hangover. OTOH I like the basic rich red of the default livery. So I figured out how to make some minimal tweaks to that, which I think are a big improvement. This one and the previous two are ready to roll. I just have to do the mips and pack them.

Then just for the heck of it, while I was on a roll, I roughed out a blue one and a brown one. These aren't bad, and may be worth finishing at some point.

*(actually should be BR 01, which stands for Baureihe 01, but whatever)

[Just Crazy Jim]

All of this may be considered off-topic, so I'm marking it as such.

Off-Topic:

TBH, I have always preferred the Vader livery on the bulk of European/continental steam. It's a purely personal preference, no doubt brought about by my early experiences seeing late steamers vetted in grit, soot, rust, and grease thick enough to be paint. I am pretty sure I didn't see a "show horse" livery on a diesel IRL until I was well into my 30s. Don't mistake my meaning, I like the colorful liveries, it just doesn't sit well with me to have every locomotive on the map tricked out like contestants during the evening gown portion of a beauty pageant.

About the skins being made at 2048x2048, that's probably conservative. I've seen video of the production line in a variety of film and game studio digital sweatshops and based on what was on the monitors in the background, I'd guess they use 4096x4096 and upwards. Which is lovely if you happen to have a supermonitor that will allow you to see it at 1:1 pixel resolution, but rather impractical for us poor slobs that are stuck with normal commercial monitors that cap out with a 1080 pixel height. I just did a quick search of monitors on three sites and the highest resolution I found was 5120 x 2880 for a price equal to about three house loan payments, and even then it wasn't new. Rather impractical, indeed.

But, back on point: When I am making a texture from scratch, the first thing I do is make a 2048x2048 and start selling off the blank acreage to the UVmap. I would like to believe it's a process I do well, but the paint on the wall would probably tell you otherwise. I learned "bigger is better" when I was first doing mods for The Sims (original) that were 256x256 bmps in 256 colors. Trying to make anything look good at that scale was nigh on impossible, so I went bigger and bigger until I was was doing everything at 1024x1024 then resizing and reducing color count. It took some time to figure out how to keep the loss from eating all the details, but I eventually came up with a method that worked for me. By the time The Sims 2 and Neverwinter Nights came along, I had graduated to 2048x2048 for workspace, then reducing as required for injection into the games.

Photoshop is a powerful tool, but it takes a terrible long time to "git gud" with it. Even the Photoshop gurus you see on YouTube (etc) are only good at those parts that they are good with and generally have no idea how to do the rest of it. One good trick I did pick up from a PS guru was using shape tool in "vector" mode to get certain elements of a texture "filled", rather than the paint bucket tool, which is always "pixel" mode. This prevents feathering when you resize the image in such instances as you want to retain a crisp line without using "nearest neighbour" on the resize. Very handy for racing stripes

Having not had much success modelling a Garratt, I think I will make a Nord Superpacific into a 4-6-6-4 in Darth Vader livery just for the giggles of it.

[Gumboots]

About the skins being made at 2048x2048, that's probably conservative. I've seen video of the production line in a variety of film and game studio digital sweatshops and based on what was on the monitors in the background, I'd guess they use 4096x4096 and upwards. Which is lovely if you happen to have a supermonitor that will allow you to see it at 1:1 pixel resolution, but rather impractical for us poor slobs that are stuck with normal commercial monitors that cap out with a 1080 pixel height.

I only have 1200 screen height, but when modding default skins I often find myself working at 200 or 400% zoom anyway. So being able to work at that scale without it being all blurry would be total luxury even if I can't see the entire skin. When I want to see the whole lot I can just zoom out.

Edit: Incidentally, how densely do you find that you can pack islands without getting problems with them bleeding into each other at a still-visible level? Obviously by the time it's stepped down to an F or G skin it's all a blur anyway, but B comes in pretty close and C isn't that far behind. I've been working on the assumption that I should generally have an 8px gap on the A skin, on the basis that it should still be clean divisions as far down as the D skin, and past that it probably doesn't matter.

One good trick I did pick up from a PS guru was using shape tool in "vector" mode to get certain elements of a texture "filled", rather than the paint bucket tool, which is always "pixel" mode. This prevents feathering when you resize the image in such instances as you want to retain a crisp line without using "nearest neighbour" on the resize.

Vector mode is great. I'll have to learn more tricks for that. I only recently discovered the difference between fill and opacity too. That's awesome once you know about it.

Having not had much success modelling a Garratt, I think I will make a Nord Superpacific into a 4-6-6-4 in Darth Vader livery just for the giggles of it.

Why was a Garratt problematic? And your proposed project sounds just as difficult.

[Just Crazy Jim]

I only have 1200 screen height, but when modding default skins I often find myself working at 200 or 400% zoom anyway. So being able to work at that scale without it being all blurry would be total luxury even if I can't see the entire skin. When I want to see the whole lot I can just zoom out.

Edit: Incidentally, how densely do you find that you can pack islands without getting problems with them bleeding into each other at a still-visible level? Obviously by the time it's stepped down to an F or G skin it's all a blur anyway, but B comes in pretty close and C isn't that far behind. I've been working on the assumption that I should generally have an 8px gap on the A skin, on the basis that it should still be clean divisions as far down as the D skin, and past that it probably doesn't matter.

What I do is based on what size I am starting the work on. But more importantly, at 1024x1024, I try to plot out the use of the total area by dividing the texture into 128 pixel blocks with an eye towards those elements that will turn to muddy filth. I don't think there's much use to even consider what a texture will look like at 32x32 or 16x16, but if you get your zoning worked out so something red stays red and something white stays white at those sizes, you're ahead of the rest of the pack. The gaps I like to keep to 4 pixels at 1024x1024, that assures that there is a gap at 512x512 and 256x256, then it gives one last crisp line at 128x128 (with no gap). If I have laid out my grid right, that line may stay crisp all the way down the chain. If something is 8x8 on it will be 1x1 by the 128x128 mipmap. To remind myself of that while working, I use a custom pattern fill as the background (see image).

On that pattern, I have a quick way to select pixels in blocks that I can be sure will look relatively crisp until 128x128. Then, for 64x64 and smaller, I rely on the power of Photoshop to resize images, playing with settings on the resize tool until I have as best as can be made of postage stamp or smaller. But honestly, mipmaps smaller than 64x64 are almost always garbage, so I don't work overly hard to make them look good.

The other end of the equation is how the UV map will scale. At and below 16x16 it's all rammed together and nothing good will come of it, so fighting to make it look good is really a waste of time.

Vector mode is great. I'll have to learn more tricks for that. I only recently discovered the difference between fill and opacity too. That's awesome once you know about it.

Gaussian blur + opacity = great shading. And usually you can turn out a better appearance of a curved surface using gradient tool + Gaussian blur + opacity than using the embossed effect. I usually reserve the embossed effect for small items like rivets and hinges. Also, outer glow set to a dark colour and multiply can give the appearance of depth on edges, like folded metal along a rivet-line.

Having not had much success modelling a Garratt, I think I will make a Nord Superpacific into a 4-6-6-4 in Darth Vader livery just for the giggles of it.

Why was a Garratt problematic? And your proposed project sounds just as difficult.

Picture, if you will, a newbie trying to get two whole locomotives' worth of drivers and bogies to new positions and then getting them to not do "the bad thing". Now add to that the fact that I rather stupidly tried to use the tender as the rear "truck". Until I have better understanding of the finicky nature of shifting drivers, bogies, etc, I will stick to easier tasks, like remodeling the cab and boiler.

[Gumboots]

That's a great idea with the PO2 grid. I hadn't thought of that, and will definitely use it.

But I was thinking of areas where you're packing in all sorts of bits and pieces, which won't necessarily fit a groovy grid. I figure those bits and pieces are probably safe with an 8px gap on a 1024 skin. But using an even 4 where applicable will save quite a bit of space.

Thanks for the shading tips too. I knew about outer glow, but not the others. BTW, I found this trick by myself: Double lining & vectors trix

Picture, if you will, a newbie trying to get two whole locomotives' worth of drivers and bogies to new positions and then getting them to not do "the bad thing". Now add to that the fact that I rather stupidly tried to use the tender as the rear "truck".

Actually using the tender as the rear truck is something I have been meaning to test live. I have a fair idea how it will behave, and it's really the only way of getting the biggest Garratts to work reasonably well over lumpy track, while keeping a consistent modelling scale for all locos (10" = 1 modelling app unit). It also has the advantage that the ability to call an extra image should make skinning easier.

The only catch is that it would mean no double-heading of really big Garratts, but they're so long anyway that having two of them in front of an RT3 8 car train would just look ridiculous.

[Just Crazy Jim]

And I am sure I have mentioned this in another post, but when you are finished with the texture for each size (made all of them) just before you do the TGA to DDS conversion, it's best to use the add noise filter (I use 1.6%) to bamboozle the DXT conversion and keep it from gobbing things up too much. I don't use the add noise filter before resizing, because it tends to get lost in the resize. By doing it afterwards, I knows it's there doing the Lord's work to keep the DXT filter from destroying my details. And, as previously stated, I don't bother with that sort of finishing touch on the smaller mipmaps, it's just wasted effort.

[Gumboots]

I never use TGA. I just go straight from PSD to DDS. But I'll try the trick with the noise and see if it helps.

Edit: Just tried a quick and obvious test. Made a basic 512x512 text image with a black to white gradient across it. Saved straight to DDS. Made another one, added 1.6% noise, saved to DDS. The two resulting images appear identical. Exact same banding and aberration in both.

Made another image. Same size, but this time with a sky gradient). Saved one straight to DDS. Added 1.6% noise to the other, then saved to DDS. Magic! Minimal banding and aberration.

Ok, then. How much noise is needed? Reduced it to 1%. No visible difference. Just as good as the 1.6%. So reduced it to 0.5%. Still fine. Just as good as 1.6%. Pic attached, showing the four results. Pic is PNG to give best comparison.