Sounds like Terraria v1.3.4.x broke RomTerraria again.
If I had to make a guess, I'd say that the new render pathway for water changed the number of RenderTarget2D's in the code. If that is the case, it's probably a 20-30 minute fix.
That said, I haven't had a chance to look at it because I've been working, and I won't be getting home from work until late tonight.
I'll take a look at it after dinner tonight.
Update 11/16/2016, 8:23pm: I've got the resolution patch working, but the new water effects are causing the following crash that I need to fix first...
...removed for space...
Update 11/22/2016, 11:10am: Patch appears to be working, but I have to finish my test pass before I release it. Hopefully tonight after dinner.
Update 11/22/2016, 6:41pm: I found a crash bug with 188.8.131.52. Working on patching it now.
Update 11/22/2016, 8:24pm: Okay, I've got a choice. I can either release a version that will only work with all the visual bells and whistles turned off, or I can hold back and release it once I figure out what is causing the wave mask to get corrupted and crash approximately ten frames into rendering. Right now, I'm planning on holding back until I get this tracked down.
I know that isn't what people want to hear, but I'd like you to remember two things. First, the reason that RomTerraria works as well as it does across multiple versions of the game is that I don't do source decompiling/recompiling. I do MSIL-level opcode patching. This requires a lot more time to do right because if I do it wrong, I create invalid code and that breaks everything else.
Second, the reason that you never see anything about this mod over on anything official is because back when Terraria launched, some asshats were using my mod as a crack for Terraria and they bundled my mod with the cracked version. Because of this, Re-Logic appears to see this mod as something bad. As someone who spent a decade in the games industry fighting pirates, I can't blame them. I'm absolutely certain that they see me as someone who has enabled bad behavior and they don't want to enable me to succeed in this venture.
I've been maintaining this mod as a spare-time project for over five years now, and it wears on me. I don't make any money off of it, partly because I don't want to further enrage Re-Logic, and partly because it doesn't seem right to me to make money off of mods. However, that means that this mod is secondary to other projects at both work and home. Even tonight, I've had to spend time helping out a lady friend.
I spend the time that I can afford to on this, both from a financial point of view, but also a mental point of view. Working on this mod isn't as simple as hitting F5 in Visual Studio and seeing what happens. It requires a lot of time to trace through each failure. It requires a lot of effort to inject diagnostics code into the assemblies. It requires a lot of patience to track down crashes, support comments, etc. I have a finite amount of willpower to get through each day and this mod gets the last few minutes of the day usually.
I'll keep working until I head to bed tonight, but I won't be able to work on it again until Thursday night or Friday morning at the earliest due to the holiday and social obligations. I'm off Friday, however, so hopefully I can refocus my energies for a bit and get this fixed.
Update 11/22/2016, 8:50pm: Around 40 frames in, something is corrupting Terraria.GameContent.Liquid.LiquidRenderer._waveMask and changing it from an array of 40000 XNA Color structs to an array of 39936 XNA Color structs. This is going to take a lot longer to figure out since Terraria uses unsafe code to work with that array.
Update 11/22/2016, 9:05pm: Well, the good news is that I figured out the root cause. There's a bad bug in LiquidRenderer.InternalPrepareDraw when it's dealing with a fixed pointer to _waveMask. This code has a lot of issues. It's not cache coherent (looping over Y inside of a loop over X), and the root cause is a chunk of code that starts writing before the variable if you are on a screen with certain conditions set:
*(vISCOSITYMASK - 200 * sizeof(Color)) = *vISCOSITYMASK;
They're trying to set the pixel above a point to the value of that point. That line should scare you. Scares the hell out of me. Until I can trace through what the hell led to this, or Re-Logic writes a cache coherent version of this method that doesn't write to memory before vISCOSITYMASK pointer, there will not be an update to RomTerraria.
Update 11/26/2016, 12:57am: A test version that seems to work with 184.108.40.206 is available over at the update post.