I want to store images and other documents in a PostgreSQL table, along with a thumbnail of each image. The original document and the thumbnail would be two separate bytea fields. PostgreSQL is running on Linux.
Because the image data could come from several different applications, I'd like to have the image processing code (for creating the thumbnail) within PostgreSQL as a function, rather than each individual application having to create the thumbnail. Is there any way for PostgreSQL to be able to create a thumbnail of an image?
PostPic sounds like the Postgres extension you are looking for.
As described in their wiki you will be able to resize and create thumbnails with ease:
FUNCTION thumbnail(i image, size INT) RETURNS image
FUNCTION resize(i image, w INT, h INT) RETURNS image
May I suggest instead that all your applications instead use a common interface or an API?
For my photography platform, I have an Upload API that everything goes through, although there are about 4 different ways to actually perform an upload (browser, desktop, phone, and software plugin). The Upload API then has the functionality to manipulate the images with some powerful and performant libraries (I'm using Python, so PIL), and then save them to the database (actually, I'm saving to a file system and referencing them in the DB, but the idea is the same).
An alternative is that a thumbnail generator service could reside outside of your database, and then occasionally loop through all your rows that don't yet have a thumbnail generated, generate one, and then store it back into Postgres.
You're asking for a world of performance hurt if you do end up doing image manipulation inside of Postgres, particularly on the memory side.
Is there any way for PostgreSQL to be
able to create a thumbnail of an
image?
No. PostgreSQL is a database engine, it just allows to store and retrieve data, and to some extent manipulate it. But doing some image processing inside it would go way too far.
Image resizing should be done outside the database.
And, as other commenter says, consider also the option of not storing the image data inside the database - only some path or locator. This is optional, but frequently it's more practical.
Read some related questions:
Storing Images in DB - Yea or Nay? ,
Storing a small number of images: blob or fs?
I've only ever hacked a few trivial functions in perl, but chances are there are plenty of appropriate libraries if you install pl/perlu.
If pl/perl2 is not an option, configure pl/perl accordingly:
plperl.use_strict = true
plperl.on_init = 'use stuff1; use stuff2;'
Simplest answer: Do not store images in database. It's slow, ineffective, doesn't scale, makes backups take longer.
And when you'll store them on filesystem - just add middleware to resize them, or a simple daemon which will resize all new images.
Related
I'm currently building my first real project that includes Express and MongoDB. Since it's one of the first backend-heavy projects I've worked on outside of my Udemy course, I've run into a lot of questions.
My project is supposed to be a mock-online store that would display items I have created inside of my MongoDB server. The problem I'm having is that I don't know the proper way of serving those image files that should be associated with each item (such as the image of a hat, for a hat item). I could add them directly into the project's public folder, but I don't know if that would be feasible in terms of the scalability that I want this project to demonstrate. But it doesn't seem like MongoDB will let me store images within each item. How would I go about doing that?
Sorry in advance if any of this is unclear, it's my first time posting as well. I'll try and provide more information if I need too. Thanks!
If you want a scalable solution for images, you typically would use a separate service like AWS S3 or Imgix.
There are several benefits to using a 3rd party service. You don't bog down your web server with image requests, or image resizing. You get virtually unlimited space. Etc.
In your MongoDB document, you would then store a key like /item/1.jpg or whatever, rather than the image itself. Your front-end then uses the key to request the image when someone visits your website.
If you want a turn-key solution, I recommend starting with Imgix (or Cloudinary, or some similar service). It is more expensive than S3, but it is pretty cheap for a small project, and it will get you up and running a lot faster.
I trying to figure the best way to store an image in a database using typeorm. Should you store it as a data url or as a buffer? Currently it would be over kill for my application to store it in a CDN hence the reason I want to store it in the db.
My thought was to save it as a data url as well as a field for the image name.
Appreciate any into!
It really depends much on your production infrastructure. By the way, IMHO the best way would consist in storing at least the file path where the content gets uploaded to (local, bucket, etc..) and the file's mimetype.
I planning on building an app whose main content are images. Basically, it is going to have multiple menus using UITableViews, whose cells are going to have only an image. When you click the cell, you'll be pushed to a simple view with that image and another one, wich has the rest of the detailed content.
This is all quite easy to do, my questions is about optimization. It's gonna have LOTs of content (Maybe 1k rows) and It's gonna display images in the UITableView, so Core Data is a must (given it's lazy loading and several other optimizations)
My question is: What's best, to store the image in the Core Data db (as NSData) or to just store the name of the image? What I'm imagining is if I store the name of the resource, For each row in the UITableView the device must go fetch that image, process it finally display it. When scrolling trough them (wich is expected to happen A LOT) we would have lots of fetching images. If I store them in Core Data, it would be as simple as taking that info and using it as if it where an image.
The benefits of storing the images in Core Data comes with the normal withdraws of storing blobs in a db. I don't know how much of a problem this would be in Core Data (My experience in dbs comes mainly from MySQL)
On the other hand, tough my "common sense" dictates saving just the name and fetching the images as they're needed its gonna take more time if they're requested more, I'm not sure how much of a performance hit would this be. Is there a "best way" to store them? Just the name and then call pathForResourse:ofType:or (if it's faster) pathForResourse:ofType:inDirectory: on the mainBundle, store the URI, or other form of pointing to it.
edit: The application will have static content shipped with the application and the user won't be able to modify this content in any way. (at least in version 1.0)
From the Core Data Release Notes for iOS v5.0:
Small data values like image thumbnails may be efficiently stored in a database, but large photos or other media are best handled directly by the file system. You can now specify that the value of a managed object attribute may be stored as an external record—see setAllowsExternalBinaryDataStorage:. When enabled, Core Data heuristically decides on a per-value basis if it should save the data directly in the database or store a URI to a separate file which it manages for you. You cannot query based on the contents of a binary data property if you use this option.
The setAllowsExternalBinaryDataStorage: essentially does what you described "...just store the name of the image..."
Also refer to these other questions:
CoreData : store images to DB or not?
Core data images from desktop to iphone
Provide example for why it is not advisable to store images in CoreData?
you will get a great optimization from just using the name of the file...
if you re-use a file ... you wont have to store it in the database twice, or alternately have a model object just to represent that file as a relationship.
You could do some profiling and check... but I assume that just a name would be ideal.
also you could do a little work to make something similar to -imageNamed that caches the images for you, and you will only have to make one UIImage for each file, wherever it exists in your program.
Don't save the images in core data. You can save the information pertaining the images in an organized matter in core data, but keep the images ordered in a supporting files section in your project. Or if you are downloading the images, you can cache them in your images section of the app and simply update the information for the images in core data.
So I am working on an app that uses core data to store attributes of objects that the user can set. I have primarily been focusing on the first part, which was setting everything up on the core data side, and now I am ready to move on to the next part which deals with the camera. I want to be able to add a city, or other geographical location (this is the object) and then be able to take pictures within my application that would accordingly store these pictures (or attributes) under the object. When I click on New York on my uitableview for example, I'd be able to view any number of pics that I took in NY. When I click on Florida, I'd be able to view any pics I took in Florida. So, my question is about whether or not that is a possibility? Sorry, that was a vague question. If I were to take pictures from within my app, would I be able to save them there as well? And make a list of them in my app? I guess I am envisioning a seperate core data-esque looking uitableview withing the place object letting you manage pictures so that they are directly viewable in like a slide show mode where I'd be able to see all the pictures I took at that location. Is that a possibility or would they save in the default picture location in the pictures app? Would there conceivably be any memory restrictions? Or would I be able to pull from the media memory (8gb, 16gb, etc)? I guess my questions aren't for specific code examples as much as they are for directional purposes. If anyone has any insights/ideas to help me out it would be greatly appreciated. Thanks!
The design you describe is definitely possible.
You can give users the opportunity to select a picture (and/or video) from their photo library and/or take a new one with the camera and then use/edit/store it in your application. Check out the UIImagePickerController and the UIImagePickerControllerDelegate protocol.
You can store images directly in Core Data as a transformable attribute by converting them to NSData using UIImagePNGRepresentation() or UIImageJPEGRepresentation(). If you do this, it is recommended that you create a separate entity with the image data attribute so that you can load your other attributes quickly (and only pull in the image data across the relationship as needed). Alternatively, you can store the images in files in your app's Documents directory and save the filename as an attribute in your entity.
Check out the TTThumbsViewController in the Three20 library for a nice way to show a grid of photo thumbnails.
I would really recommend storing the files to the application Documents directory, and storing paths to them in the database. You can do it the other way but storing them in files keeps the database smaller and gives you more flexibility to upload them somewhere without having to load the whole image in memory.
I want to create and manage a database with images and or audio clips. I know it's not the best idea and I know there are better options, but it's the easiest way to have the data separate from the programming and I'm not the one writing the code. So I need an application that will allow me to edit the database that the application is calling, and the application needs to call random audio or image file. I JUST NEED TO KNOW OF AN SQL EDITOR.
Super bad idea to put large files in a database, it will kill performance and could well blow away the memory limits an application has. You also cannot stream them out of the database the way you can from the file system.
Instead, consider this approach - as Alex suggested, work with separate files and strings that represent file names. You could put both the database and the files into a single directory, that your coder keeps as a folder reference in XCode - so all content in the folder you change is added to the coding project automatically. Note that if you change any existing file, due to a bug in XCode he'll be required to do a Clean BUild before building again or it will not copy changed assets (mostly a problem for the DB).
Then you can easily use any SQLLite client to maintain the database of filenames and other data. "Base" is a nice standalone app.
I would highly recommend you use some source control system like Git so that you could check in changes and the programmer could get his project updated right away without the confusion of emailing files around.
You could use Core Data instead. It uses a SQLite backend, by default. Instead of using BLOBs, you can simply store an NSString* that is a path to the object in the application's Documents folder. When you want to retrieve a stored image or other large binary data object, you can load an NSData* instance from the path value directly. Keeping large files outside the database will give you much better performance.