What’s new in Adobe Experience Manager Assets as a Cloud Service

Transcript

Hi, everyone. My name is Alex Tiers. With me today is Chandar Natarajan. I’m a Senior Product Manager for the AEM Assets team and Chandar’s Senior Engineering Manager, also for AEM Assets. And today we wanted to spend some time going over some recently added features to AEM Assets Cloud Service. Really exciting things being added every month. And we wanted to cover these developments to give you a heads up on these features and give you some details to implement best practices for your customers. The agenda we have today is a little bit of background on why we’re holding this session. We’ll give you some updates on metadata schema, a new bulk import tool we’ve added, as well as the exciting add-on to Cloud Service content automation, which is the Creative Cloud APIs integrated directly into AEM. And finally, we’ll wrap up with a live demo. And Chandar, you want to kick it off? Sure. This session is about the new features that we added recently and how it adds value to our customers and in return what we get on investment. To start with, metadata schema. So we recently added most often request metadata schemas in the Cloud Service. One is a checkbox and the next one is the text area widget. And last but not least is the root path selector for the tags widget. So now, as a customer, we could add these metadata schemas to enable your metadata with the right property for viewing it after updating it for the customer. And the root path selector for the tags enables you to have multiple tag widgets on your schema and also to help get rid of some of the technical tags. Next is on bulk import tool. Bulk import tool enables customers to upload or import thousands of assets from their cloud. For example, if you have a large retail customer who are trying to upload thousands of assets to prepare for a holiday sales event, they could use this tool and then do upload it to Cloud Service. And then do uploads on a periodic basis. And this is widely adopted tool by multiple customers. We released this a few months ago, but it’s been widely adopted. And for this, we added a few more enhancements to make this tool more usable. Just one is the scheduling feature. So now customers can… Sorry, let’s please… To start with the scheduling feature, now customers can edit on hourly, daily, and weekly basis using the same job configuration. For example, I mentioned if there’s a holiday event coming up and the customer wants to… customers plan to get assets from their sources on a daily basis, you can configure it once and ask them when the new assets are uploaded. Yeah, this cron job will automatically kick off and start uploading it on the scheduled time. It can go as low as hourly basis for this feature. Next, additional configuration support for the bulk import tool. One is now we can have a regular expression to filter, to include or exclude assets from the cloud storage. Again, for the example I shared, you want to only pick the assets that’s particular… that’s pertaining to that holiday event. You can write a regular expression to pick only those assets. Or if you want to exclude some special files or you do not want to include as part of the import, you can also specify that regular expression on that text field. Next is more on cost savings. After you import, if you want to remove those files from the cloud storage to save cost on the block storage, you could check that box to delete after import. Once the bulk import tool successfully imports the asset, it will remove from the source. Last but not the least, we also have support for adding metadata. You can specify a CSV file, which should co-locate in the cloud storage as the source to upload metadata with the assets. When you upload bulk import assets, you could also specify a CSV and place it next to the source. The bulk import tool will import both together. The format of CSV is the same as the existing metadata import functionality. Next is more on enforcing or hinting on a best practice that we all know about the folders, the limit on number of child nodes under a folder, which is 1,000 per folder. So the bulk import tool, when you try to import a lot of assets, it’ll check, and if there is a violation of this best practice, it’ll show a warning both in dry and the regular one to notify or hint the customer not to proceed with that. And it logs it and shows immediately in UI processes within 30 seconds, it will show that warning to the customers. That, I will hand it off to Alex Thiers to share more cool features that we added in content automation. Thanks, Jhanda. Great stuff. So now I’d like to talk about content automation, which, as I mentioned earlier, is a recently added add-on for AEM Assets Cloud Service, which integrates various creative cloud APIs directly into AEM. Now, we’re using processing profiles to apply these different operations at scale. And some examples are things like Lightroom presets, swapping smart objects, and even applying Photoshop actions in bulk at an entire folder tree of content via these processing profiles. And this unlocks a number of capabilities that truly accelerate content velocity. Whether you’re retouching images that are coming directly from a photo shoot with Auto Tone or the Lightroom presets, or you’re generating marketing-ready assets like banners or product imagery, you can process these natively within AEM without ever having to go to the Creative Cloud applications. This allows you to speed up the ever-increasing demand for content and really speed up your own content operations to meet that need. Now, let’s have a little bit more of a look what those operations look like. So, starting at the top, Photoshop actions can be thought of as a macro. In the Photoshop desktop application, what you might do for a repeatable task, something like cropping or resizing an image, adding a layer with a standardized name, some of these basic Photoshop operations that are repetitive, you can record this action, which, like I mentioned, is sort of like a macro recording. And moving forward, you can play that recorded action directly within Photoshop to apply to new assets. Well, once you’ve created that action file, you can save it and upload to AEM. And once you’ve defined your processing profile with that action, you can then apply that profile to a folder, just like you would for generating out-of-the-box renditions or generating new video encodings. The assets that you upload to that folder with that profile applied will automatically be applied to that action. And in this case, you can see we’re adding not only a promotional banner here, an overlay, but we’re also doing some significant artistic adjustment of that image. Next, we have smart object replacement. And what this does is takes a Photoshop file with a layer that’s designated as a smart object layer. And the cool thing about smart object layers is that they actually have built-in support for vector files. So in this example, we see a model with a T-shirt. And it’s expected that the logos that are being applied to that T-shirt are vector files, could be AI files or another format. And that smart object layer will preserve the scaling of that vector file in that layer itself. Really handy for swapping out different icons or different logos while keeping the rest of the image the same. And with that in mind, we can set up another processing profile that has that PSD file defined as a template, so to speak. And you can define a specific layer to swap out. And again, once that profile is applied to a folder, every single image that gets uploaded will be applied to that PSD template according to the profile definition. Really, really cool way to automatically generate new product variations. And as we’ll see later, you can even generate new banners or new travel marketing content, a whole variety of content, not just for product imagery. Next, we have the ability to auto mask and drop background. Now these are kind of two sides of the same feature here. So what the mask is doing is auto is using an AI based feature to automatically determine the contrast and determine what the foreground subject is and what the background is. And by generating that mask, which we see here, we can also do an automatic background knockout here. What’s cool about this feature is that you can get both of those outputs, both the image with the background knocked out as well as the mask to use downstream or to touch up additionally in Photoshop with AssetLink or CC libraries as well. Within Lightroom presets, again, similar to Photoshop actions, you can define this preset and apply it in AEM assets with the same processing profile. Now Lightroom presets can be anything from a stylized image treatment to auto toning something that you know needs a specific color gradation, color profile, so to speak, as part of that processing workflow. You can also really quickly iterate on new ideas with this way, like generating new variations of existing marketing content, say for seasonal releases, or a promotional campaign that may have a different color theme that’s not the typical brand colors. You can use these Lightroom presets to create these new variations and automatically generate the resulting renditions from your existing assets. And finally, we have some additional AI based features that both auto straighten images based on the determined horizon, as well as auto tone them if in case the exposure and light balances is not quite correct. Again, great example is uploading straight from a photo shoot and allowing AEM to automatically adjust that imagery for further downstream processes, whether that’s to autocrop and then publish or even run through another content automation workflow like we see here. So let’s see what this looks like. Now here I have my AEM cloud service window open. And I have a folder here, content automation demo folder where I’ve set up each of these operations. So let’s go let’s let’s do a few of these live. And while some of these assets that I upload are processing, I’ll talk through a few of the others. So first off, let’s go check out image cutout. Now, you may recognize these headshots of Chander and I from the slide. And actually, I used AEM here to generate those cutout versions of our headshots. And when we upload them, there’s the standard out of the box renditions that are generated. We can see here the original JPEG. And we can also see that we get things like the web rendition, a zoom rendition, as well as some thumbnails. But we also get this cutout rendition here. Now, I haven’t defined a specific size for this rendition. So the cutout operation is going to return a result that’s the same size as the original asset, the same dimensions. And what we can see here is that the service did an excellent job in determining the difference between the chair and I’ll toggle back to the original. And you can see the chair headrest in this black and white version. I would expect to see. Yeah, yeah. Chander, go ahead. We’re still seeing the slide if you’re sharing something else. Thank you very much. Nope, appreciate it. Let’s share our window now. Thanks, Chander. All right. There we are. So let me just back out here so we can see a little bit better here. So here we have the images that I was referring to, which we uploaded in the slide for the session here. And we can see both of these are just the original images. And when I click in and look at the rendition drop down here, we can see the different renditions that are generated out of the box. We have the original here, a JPEG, and we have some other additional out of the box renditions that are already defined. A web rendition, a zoom, as well as some thumbnails. But here we can see the cutout. And this processes within actually 10 seconds or less in most cases. In some extreme high resolution images, we see a little bit longer processing, maybe 20 to 30 seconds per asset. But by generating this cutout rendition, the service has automatically dropped out not only the background, but also the headrest of Chander’s chair here. And we get a really nice usable rendition that’s included here with the out of the box renditions. And Chander, apologies for zooming in. But what I find so impactful about this is the handling of human features like the hair of a person, which typically for anyone who’s done creative production, it’s such a challenging process to do. Having done some creative production myself, I can honestly say that cropping out people is my least favorite thing to do. And it just blows me away when I look at the results of this service, how perfect the results tend to be on what I consider one of the hardest use cases here, which is human hair, really incredible performance, not only in the results and the usability of those results, but also the sheer speed in that they come back. And in some of our customer cases, we’ve seen this alone show the opportunity to accelerate their velocity of processing images and processing headshots by 50 times so that the time that it would typically take them to do one image, they can now do 50. Let’s upload a couple others. I have another headshot from our moderator here in June, as well as some shoe assets that our 3D team just rendered out and shared with us. So to show you that the service can handle not just headshots, but also products, and not just products and silos, but real lifestyle imagery like this. Let’s see, we just got this up. So we’ll give this a few seconds to process, but we can see here what we’re uploading. Now, what I want to call attention to here is the sheer speed in which these assets process with the Photoshop APIs. We’ll see the status banner here. We’ll stay in processing for a little bit longer, because what’s actually happening is we’re running a number of processes, including smart tagging to generate our metadata tags. But when I jump in here, let’s see, this is really just uploaded a few seconds ago, we already see not just the out of the box renditions, but we already see this cutout rendition in a matter of less than 10 seconds. Again, I would say this is 99% perfect. If I’m being really, really harsh on the results, I would say I could see a little shadow here on the right hand side. But overall, I would say this is near perfect. Again, in a matter of seconds. Now, let’s see what the shoes are looking like. Again, we already have the cutout rendition along with all of our other out of the box renditions. And here again, we can see near perfect results. So really exciting when I think one of the things I think is most exciting about this image cutout operation is that it’s so easy to plug into existing customer workflows. Functionally, it’s a pretty straightforward feature. It’s dropping out the background. But in terms of the repeatability of this operation, it’s something that I think all organizations that have image creation workflows, they struggle with at one point or another, masking and knocking out backgrounds, and the ability to just completely automate that directly within the AEM. And preventing the need for those professionals to even go into Photoshop at all, really has a lot of potential to accelerate. Now, let’s have a look at another one. I’ve got here another folder with the AEM weekend banner set up. And some of you may have recognized this already. We had a vertical banner here that our creative team provided. And in this example, this is a PSD file. And the background image is a smart object layer. This nice landscape image. And what I’ve done is I’ve set up a processing profile with this image. But you know, my creative team actually just provided a new version of this. This the first one was a portrait version. But here we have a nice landscape version as well. So what I’m going to do is show you how to update that processing profile. And then we’ll add some new images here and see the result of not one, but both of these new banners we’re going to create. So I’m going to navigate over to my tools section, down to assets. And I’ll select processing profiles. Now, once content automation is enabled for your customer, remember, this is an add on, so that needs to be licensed on top of the cloud service license. Once it’s enabled, you’ll see the creative tab here in processing profile. And this unlocks that whole suite of operations, which we can see here, lightroom preset, straighten, auto tone, cut out mask, smart object and applying actions. And what we’re going to do here is we’re going to be replacing that smart object layer. Now, I have both of these set up. So what we’re doing here is we’re going to first select our creative operation, which is the Photoshop smart object replacement. Next, we’re going to select the PSD file that we want to act as that template. And finally, that file when it’s uploaded will be parsed and we’ll get a return of all of the layers that are available. With that smart object characteristic. And here, there’s just one in this file. So we’ve selected that. Finally, I’m naming what I want that rendition name to be called, as well as defining the extension and other characteristics of that image. And in this case, I just want it to be a full resolution PNG. Same down here. Now, this is that new landscape PSD that my team sent me. And same thing. I’ve selected that same layer. And I’ve adjusted the naming of that rendition so I could tell the difference between the portrait and the landscape. Everything else I’ve saved the same. And you’ll notice here, this is a single processing profile, but with two rendition definitions inside of it. So we’ll go ahead and save this. Now let’s go back to our folder. Let’s go ahead and upload a few new images to this folder. And we can see what results we get. I’ll also jump in and show you some of the existing ones we’ve processed to see how this is handled. So here we have our weekend folder. And I’m going to just select a few of these. Some nice outdoor images. Just drag and drop directly into this folder here. Now, the benefit of the processing profile approach is that once configured, this asset processing is part of all of the other out of the box operations. There’s no need to set custom workflows, or to set up new event triggers in order to implement this feature. Of course, you can set up things like approval workflows to notify certain users when these are created to approve them, for example, or to review them before they’re sent to the automation service. But overall, the functionality is built such that it can be just part of all of your asset processing that happens by default. Now we’ve uploaded these images here. I’ll give it a little bit more time to process because to be honest, the image cutout is incredibly performant. With this with the smart object replacement, and also I find with the Photoshop actions, these are a little bit more complex in that they’re handling really the the insides of that Photoshop file in or action file in a much more detailed way. Anecdotally, the performance that I usually see for these features is tends to be a little more like 30 seconds to a minute for end to end processing for the asset. Now keep in mind that’s processing with that API service. So you might still see these images and processing state, because they’re also being smart tagged and any other downstream workflow we have set up. So let’s reload this and I will start checking out some of the results we have here. First off, let’s see this first one that we uploaded a bit ago. Now we can see the original is a high res JPEG, fairly large. And again, we have the out of the box renditions. But here we have not only our landscape banner, we also have our portrait banner, both ready to use on any channel that needs those different orientations of renditions. Now we can see those five images I uploaded earlier, they’ve already finished processing. So let’s take a look. Now here I have my landscape banner. And that looks really good. One of the things that I really like about this is that it’s a really nice way to get a lot of information out of your landscape. So that’s really good. One of the things that I’ll call out is that with these new banners, we define the PSD file to have the specific resolution of 1920 by 1080. But you can see the original asset is not standardized like that. And that smart layer swapping will that smart object layer swapping handles that resizing perfectly. Here we can see the portrait version of this banner, again, just perfect for a downstream use. Let’s go ahead and check out one more here. Let’s see how our mysterious foggy forest looks since we are getting close to Halloween here in the States. This is a perfect Halloween themed image to use for a new seasonal banner. And just about as good as it gets. Again, I just love the performance of these services, not just in the speed at which they process and scale up to processing thousands of assets automatically, but just the sheer quality of the results I find are breathtaking every single time I run this demo. And that’s it. Thanks, everyone. Hope you enjoyed the demo.