Most people don’t care how much a developer loves/hates a device so I’m hesitant to write reviews of hardware but the iPad 3 changes that. No, I’m not going to rave how it’s better than sliced bread. No, I’m not going to whine about being a modest upgrade.

The iPad 3 feels like the iPad 2 Intended Version. Let me explain.

While we were all reading rumors of iPad 2, a few points stuck out. The iPhone 4 was released after the iPad 1 so should the iPad 2 receive the same pixel doubling? Why didn’t we in the end?

The answer is probably production related. Either the standard A5 couldn’t drive it with reasonable performance or LG/Samsung/etc couldn’t make them in high enough yields. I’m going to blame the display manufacturers. Apple could’ve made the thing iPad 1 sized if the battery and processor combination were a problem.

Whatever the story, Apple couldn’t ship iPads with retina displays so they put the standard display on it. Upon doing this, they probably discovered all their efficient new components had something like a 20 hour battery, so they shrank the battery and case down to match the existing 10 hours.

Eventually the display manufacturers produced a high yield retina display and Apple revisited the idea with their new quad core A6 processor. With the iPad 2 sized battery battery life was probably very poor so they made it as big as their conscience would allow (the current battery size). Still not getting the performance they wanted they had to ditch the immature A6 and add its GPU to the A5.

I had speculated earlier that some devices might see a die shrink A5 rather than a new processor and when I heard about the A5X I thought that was the case, but it turned out it’s the same size (architecture) as the A5. I maintain my skepticism that we’ll see quad core iDevices from Apple anytime soon unless they can nap/powerboost efficiently and instead we’ll continue to see screaming dual core CPUs with screaming GPUs.

Yes I’m sure there are way better things than DAZ and Bryce (Vue, Maya, duh) but I don’t have that kind of money right now. DAZ is cheap (free this month for some reason) and has an affordable content store. Unfortunately, All pro software (especially for the 3D nature) is very poor at working on current versions of things. Bryce doesn’t work on Lion. DAZ on Mac is clunky. Bryce didn’t work on Windows 7 64 after I installed the content. I finally decided I wasn’t doing anything productive and just gave in and made a dedicated XP Pro SP 3 virtual machine (shown above).

Installation went great and I even got a model from the store up and doing its thing. The problems at this point were minor but here’s how I addressed them.

DAZ models are light and even a 2048×2048 render takes at most a minute, so no big deal on a virtual machine, but saving them to the VM and then getting them back to a Mac is annoying. To solve this I added and SMB share on my OTHER Mini (the one the VM isn’t running on) and mapped to the VM. Unless you’re on public WiFi, if you intend virtual machines to talk to the rest of your network make sure you set them to Bridged.

Anti Aliased Bryce renders can take overnight sometimes, so rather than deal with the VM being locked up, I installed the Mac network rendering service (again, on the Mini that isn’t hosting the VM).

It’s annoying. I wish I could just do it all natively. But, at least I don’t have any problems now. Things are working. See?

I’m going to start by mentioning Apple discourages all of the above and prefers you make a free app with in app purchases. Having said that, here’s what I think happens depending on which type you use.

Using “Lite” for the free version of “Pro” for the paid version (or doing both) makes the user more likely to get the free version. After all, they’re not a professional, especially when “Pro” is applied to mean ad-less. Who in their right mind needs the “Professional” FML reader? When people hear “Pro” they’re probably going to think of desktop Photoshop, and if they don’t use that then they don’t need your “pro” app either. I’m guessing using “Pro” and “Light” are somewhat effective.

“Free” is probably the most effective simply because it makes the user feel cheap each and every time they see your app on the home screen. If they use your app long enough, their internal embarrassment might make them upgrade at some point.

“Express” is probably the worst name you could use for your free version. Express means fast. To a user that probably would’ve bought the paid version, express can mean “without bloat” or “better startup times”. If I use the express version and never know what features I’m missing I’ll be happy as a clam. So with that in mind, Express is a great idea for the name of your $0.99 app when you also sell a $4.99 or $9.99 app. You’re getting a customer, and you might get a double customer. I wouldn’t attempt applying this to games.

1 – The Lurker
The lurkers if your casual person browsing the web who’s tired of everything and anything having an instagram effect applied to it. I am usually in this category. If I never make a single buck off of it I’m still happy to have made Normalize so I can browse reddit in peace.

2 – The “Photographer”
These are the types who troll the photos category religiously and subsequently fill the web with the types of images that piss off type 1. The premise of Normalize confuses and sometimes insults these people.

3 – The casual Photographer
These people use photography apps on their iPad. Their photos come from iPhoto/Aperture syncs. Maybe it’s a DSLR, maybe it’s a 12MP point and shoot, maybe they just organize all their iPhone pics. Their only goal is only to salvage a shot that went wrong. They may be below the skill level Photoshop requires and might not understand histograms in aperture but are frustrated when the magic wand in iPhoto doesn’t do anything.

Obviously I’m not interested in type 2, but they do exist, so I need to work on my app store description. Perhaps if I change it to free with an in app purchase for some fine tuning tools I can get more 1 and 3.

There’s a reason modeling software focuses on curvy women – it all sucks. Without motion capture posing is damn near impossible. It’s so difficult that poses – just poses – for existing models are sold separately and by third parties.

And just to pour salt in the wound, these apps all have Mac versions (they’re usually written in python not a real language) but with second class support. “Mac Versions” of “older” (like 2 years ago) models come in .sit files, which decompress into a VISE (remember those?) PPC .app. Newer ones are zips but still an installer executable (not a bundle.app). The Windows versions are .zips of .exes. Why can’t they just be a single .zip with a few shell scripts?

This weekend I was using my MacBook Air 11″ to remotely create a virtual machine on my HTPC Mac Mini because Bryce doesn’t work on Lion and I need to finally get some stuff done. In the process, I had the remote display set to 1920×1080 and viewed it Full Screen on the Air’s 1366×768. As both are 16:9 there were no borders but one thing I did notice was that despite the obvious drop in quality, 1080p was perfectly usable (readable) at 11″. Upon declaring whatever DPI this is “usable”, I opened up an XGA sized window in the remote session and measured. To my surprise, it was actually less than 7″. In other words, it’d be almost like 4 iPhone 3GS screens glued together.

My main sticking point for why Apple can successfully make an iPad Mini is that it should be able to fit into a coat pocket or at least a purse. At $299, there are enough wives and children who currently only have limited access to “Daddy’s iPad” to sell plenty – my brother, a father of three, and our father both share a single iPad with their wives.

As a developer, of course I’m going to buy it. But realistically, I can see me asking “Honey, can you carry this in your purse?” every time we go somewhere together. Yes, an iPad 2 with smart cover fits in her Vera Bradley purse but it’s not casual. You’re very aware there’s an iPad in your purse.

Back to screen dpi… Safari’s Reader function kills whatever worries we used to have regarding the annoyance of pinching/zooming. The first generation won’t have a retina display, but can you imagine if a second generation model did, and a first generation iPad Mini without retina display was sold at $199?

Auto Adjust is a universal application and has been since the iPad came out, but I wanted to see how it would look on the iPad’s Retina display in “2X” mode.

The good news is that iPhone apps finally render at their own retina resolution. The bad news is that that’s still less than half of the iPad’s native resolution. iOS 5.1 when not on a retina display still shows these apps at 480×320.

Click for full size:

Auto Adjust running in 2X mode on a retina iPad

Although some elements are scaled using bilinear filtering, you can see the letters in the UIToolBar are nearest neighbor (pixel doubling). Checking with iDecorate showed similar inconsistencies of what gets scaled nicely and what doesn’t.

iDecorate running in 2X mode on a retina iPad

But there is one more piece of info worth noting: the background images for iDecorate are 1920×1280, and it appears the full sized image is being used rather than being scaled down to 960×640 and then scaled back up.

None of this concerns my apps because I write Universal binaries, but it matters for people who refuse to pay for HD versions (or for apps that lack Universal versions – I’m looking at you every speed test app ever).

Marco Arment’s Learning from Competition inspired me to reflect on iPhoto now the Top Paid App, period, and what it means for me and my apps. While the $1M a day it’s probably earning at that spot would be epic, I’m not after that much money. $1M/year would be epic. $100K/year would be pretty damn awesome. I’m happy as long as I can afford sushi for 2 on Saturdays.

First, I think iPhoto is a much more serious threat to things like photogene which are general purpose editors. iPhoto is a general purpose editor with more features, but with great power comes great complexity. Without watching the demo or using the in app help, I had to do a lot of poking around to figure out what exactly the app could do, and I’m still not sure. I’ve always felt that the best iOS replacement for a desktop app isn’t some complicated $5 or $10 app but rather it’s 5 or 10 $1 or Free apps that have focus. When you launch Auto Adjust, the only thing you can do is open a photo. When it opens, the next thing you see are a bunch of simple controls which then move themselves to fix your photo. You’re welcome to fiddle with them, open a new photo, save, whatever at that point. Since iPhoto is equal parts photo organizer and fixer, that concentration gets a little hazy. When you’re adjusting something, you can flick left or right to the next photo in the album. It’s not clear what happens to your changes on the other photo. Were they saved to the original, or a copy, are they still temporary, were they not saved at all, will they be there when I relaunch the app?

IPhoto’s editing features are amazing. The HDR capabilities (dimming clouds while lightening foreground) are incredible and something I wish I got around to perfecting. There are some quirks I hadn’t found a good way around yet that Apple seems perfectly OK with. Specifically, when you do crazy things to curves or color ranges, inflection points and border regions can end up looking inverted or loose all saturation. The fact that Apple does it based on colors feels like cheating to me too. I like doing things that work in all scenarios.

This will be the focus for my new features. The Auto Enhance in iPhoto seems no better than it is in the stock Photos app. There’s also so much UI in iPhoto that your photo is barely the focus of the app. This desktop-like experience is definitely what I like while wearing my Pro hat, but I’ve never considered that my demographic. I’m glad I introduced Normalize so I can go in two directions at once. It’s worth noting that adding an app didn’t split my sales. The net results is positive but less than double. You may not see updates for a while though. Being recently compiled and presupplied with 2X artwork both apps already looks great on the retina display, so I’ll forgo the “adds Retina display support” update and wait until I have new features to introduce. Also, it’s about time iDecorate got a content upgrade.

Oh, and one more thing, Auto Adjust works on iPad 1 and the original iPhone.

The other day a hack became widely available to run the iPad Simulator at 2X, (2048×1536). Perhaps surprisingly (perhaps predictable, it IS Apple after all) other then springboard icons, the Apple apps were all ready to go and rendered properly. But much more interestingly, my current shipping builds of iDecorate, Auto Adjust, and Normalize (which contain @2X artwork – I’m proactive like that) also rendered 100% accurately. No Scaling.

The reasons for this make a lot of sense. iOS could operate in 2X mode for over a year now. It just never was shipped on an iPad.

Guess what will look like shit:

All those magazine apps which are really 1024×1024 rastered images. If you think 500MB per issue is bad wait until you see what happens next. Of course, it will be a while before Adobe updates their tools to support the retina display. Perhaps if Adobe is using bastardized UIImageViews at least the content will be interpolated and not nearest-neighbor.

“Stolen” from Final Cut Pro (and everything else), my three-way color corrector wheel is possibly the most powerful and intuitive way to color issues. Yes it’s a bit of an acquired skill, but much less of one than looking at histograms. The flip side of that is that it’s not a precision tool. Either way, it can give amazing results.

The algorithm behind the wheels is surprisingly simple. From a high level, you have a location on a circle, or an angle and distance in polar coordinates. Translate that point into a hue and value and shift the channels accordingly and you’re done! One common complaint I receive about it though is that mine are too tiny.

I know.

Auto Adjust 4 is still in the design phase, as is an update to Normalize, both of which will feature a new solution to the color corrector wheels. But that’s not all. Auto Adjust 4 will make good on my promise to increases Auto Adjust’s power without dumbing down the UI (which is exactly what Normalize is).

I don’t know if the algorithms will still be printed in the final version, but for those that don’t know, there are many ways to desaturate an image. The most common method is to average the 3 channels. This often results in a lot less contrast than the original image, and going back and adjusting the levels (contrast) is just lossy at this point. So, I’m introducing 3 commonly used alternatives for computing greyscale. The most accurate I’m offering is Rec. 709 luma (Y’) which is simply a weighted average rather than a linear average.

While this will have an obvious difference if you completely desaturate your image, this setting is actually going to permutate throughout the rest of the app. What other photo operations need to know brightness data?

HDR – If you think you’ve seen good HDR apps you haven’t seen anything yet.