Webinar: Can AI Really Retouch Images at Scale?

Listen to Our Recorded Session: Experts Discuss the Role of AI in Scalable Image Retouching
Published on Aug 10, 2023 by
Rahul Bhargava

About the Webinar

Curious about how AI is shaping the future of image retouching? We recently hosted a webinar featuring our in-house experts at Crop.photo. They delved into the practicality of AI in handling bulk image retouching and addressed some common concerns around its scalability and effectiveness. If you've ever wondered about the advantages and limitations of AI-powered image editing tools, this is a must-listen session for you.

Ready to gain some insights? Click here to watch the recorded webinar.

Feel free to share the link with your colleagues and join us in reimagining the future of image retouching.

Transcript Available

For those who prefer reading or need to quickly reference key points, we've prepared a full transcript of the webinar. The transcript covers the in-depth discussion and answers to audience questions, providing an invaluable resource for anyone interested in the topic. Check it out:

Rahul        00:01:47    Hello and welcome to crop.photos, AI in e-commerce webinar from San Ramon, California in San Francisco Bay Area. I am Rahul , I'm the CTO  and founder of crop.photo. And today joining me is Joyita. She's the product lead for Crop.photo. Uh, before we begin, we just want to let you know that you can ask us question through the q and a panel in Zoom at any time, and I would actually encourage you to send us your question in advance because we'll be taking them in the order they arrive. Alright, then it's time to get started. So today we are finally going to answer the question, can AI really retouch images at scale?  
Rahul        00:02:36    So before we begin, quick look at the agenda. So we are going to first go over the typical challenges that you face when you are working with e-commerce product imagery at scale. Then we are going to talk about AI's ability to address these challenges or not. And then we will do a live demo of, uh, AI image editing tool crop photo to show you whether or not it can address the challenges that arise in e-commerce product imagery editing at scale. And then finally, after the demo, we will talk about what's in store for ai, what does the future hold, and then we'll get to the conclusions. We'll have a round of q and a. We'll also be having polls throughout the webinar. And if you wait till the end, there'll be some previews that we'll be sharing with you.  
Rahul        00:03:33    So let's talk about the challenges of e-commerce editing. So if you are involved with e-commerce, either with, uh, a small business or a large brand, these challenges are, are something that you see day in, day out. So the first one is turnaround time. So you have to work on product SKUs, new ones. Sometimes you are relisting an existing set of product SKUs no matter what. You might be dealing with 50 or hundreds of images. You might have a sale that's happening in the next few days, or you might have a catalog that needs to be updated every season. And it's a struggle to get high quality edits done in time. So that's the first challenge that we see. The second one is production capacity. And what do I mean by capacity? So you might have a team of researchers in-house, you might be outsourcing to freelancers, VA or offshore retouching firms, but either way, you are limited by the resources that you have staff at at these various uh, uh, channels.  
Rahul        00:04:41    And what happens typically in e-commerce is you might be on a cadence, you might be doing 50 or 10 images a week, and then suddenly something happens, uh, sales arrive, uh, event happens and you now need to double or sometimes quadruple the amount of images that you need to produce, uh, and, and upload that to the various, uh, marketplace and and channels. So production capacity can be a huge, uh, challenge. The final point that I want to talk about is the cost. So image editing is a huge time suck. Whether you are working with 10 image or 50 image, what we find is, uh, average vi toucher can take anywhere from few minutes to eight to nine minutes per image in order to finish the editing task. Now, even if you are outsourcing to an offshore retouching firm, you still have to deal with project management overhead.  
Rahul        00:05:36    You have to get your images to them through Dropbox or Google Drive. And then once the edits are done, then you have to review and all these operations take time. So these are some of the challenges that you see in bulk image editing. Now, before we talk about what AI can do or cannot do, let me also walk you through a typical e-commerce post-production scenario. So when you are retouching images, you're performing certain low to mid-end operations as well as high-end operations. But the fact remains that independent of the e-commerce imagery, the type of imagery could be on model images. There could be product or flatly images. You might have lifestyle photos. All of them require certain basic operations day in, day out. So you have to crop the object of impressed or product so that it looks great without any distraction. You have to often replace backgrounds, replace it with a wide background or in grad background based on where you are uploading your images to.  
Rahul        00:06:40    You have to resize these images for your marketplace dimensions. If you are a Shopify storefront owner, you might have to change the aspect ratio to two to three. If you are a Amazon seller, then you might need to create a 1600 by 1600 pixel image. If you are running your own website, then you might have certain crops that you need to to adhere to the standards, uh, maybe uh, uh, uh, varying across marketplaces. And what we find is across all these low to mid end image operations across various options, there are cost involved of course. So if you were to use offshore firms to you do your retouching, you might send them a batch of couple of hundred images and it it'll cost you, uh, uh, a decent amount of money. You might spend hundreds of dollars doing that. And that's not just the cost of the service, it's all the project management cost included as well.  
Rahul        00:07:34    If you like to roll up your sleeves and do it on your own, you might be using Photoshop to do that. And in that case, you might burn through several hours, sometimes even days to go through 200 images. And then as we look at ai, you will see the same amount of images, 200 images. You could get it done in few minutes at a fraction of the cost. And I'll talk more about, uh, uh, AI in a moment. But before I do that, let me just show share with you. Um, something here real quick. So this is an example of what not to do. This is a e-commerce store, and I don't want to demean them or anything, but it shows that somebody was in a hurry. The images needed to be uniform background, the aspect ratios needed to be the same. But you can see none of that is happening here.  
Rahul        00:08:21    There are a variety of backgrounds. Some of the objects have different type of shadows, and that's not the look that you want to give your target audience. You want to present them with a uniform look. And that's where retouching is, is so critical. So now let's talk about what can AI do and what it cannot achieve today. So currently the AI is able to take care of, like I mentioned, common tasks. So these include being able to do, uh, a smart resize. What I mean by that is, is if you have products like let's say there's a shoe and you want to align that perfectly across a listing of tens and twenties of shoes, you would be able to recognize the object in your images using ai. And then content, air aware resizing can, can absolutely do that if you have backgrounds that need to be cleaned up, replaced with a white background.  
Rahul        00:09:18    There are AI models that are optimized for e-commerce that can do that very well compared to generic AI models or generic tools that are designed to work with any kind of, uh, backgrounds. If you want to add uniform drop shadows to your flat lays, you can easily do that with ai If you want to create headless crops or unrecognizable crops of your models so that, uh, you can upload them to marketplaces that require the the phase to be headless, you can do all of those operations with the, with ai. And just by way of, uh, possibilities, let me share what, uh, some of these companies, uh, do. So for example, this is a huge marketplace and this brand needed to create headless crops of their models for the apparels that they retail. They go through like two or 3000, uh, different images across many hundreds of uh, SKUs every month.  
Rahul        00:10:14    And this is something AI can easily take care of. It can resize these image to the marketplace dimension, replace it with white background and create headless crops from, uh, full model images. Here's another example. This retailer, uh, selling, uh, craft beer needs to have flatly images of bottles cropped to exact same, uh, height and and width and fit into the same aspect ratio and essentially have the ability to align them at the same level. That's easily done with, uh, uh, ai. Here's a shoe, uh, store that needs to have all the shoe imagery aligned roughly at the same uh, level, and also have similar drop shadows. And AI has no problem handling that. Here's an example of a small store that has for the same pro skew many different styles. They might go through like 50 styles of, uh, t-shirts and they need to create four or five variants. So for just one t-shirt, uh, skew, there might be hundreds of images and AI is, is perfect for that. So these are some of the things that you can expect AI to handle, but you don't have to take my word for it. You can. Um, I'm gonna have my colleague Jota now go through a live demo to show you some of these possibilities and jota. So I'll turn over to you. Um, go ahead with, with your demo. Yeah,  
Joyita       00:11:40    Hi everyone. I'll just, uh, go ahead and share my screen now. Yeah, so let me know, uh, when you can see my screen.  
Rahul        00:11:58    Yeah, Joya, go ahead.  
Joyita       00:12:00    Yeah. Thank you. Thank you Darul. So, uh, I'll be showing how uh, you can use, uh, AI capabilities of crop photo to edit your images in Val. Uh, so, uh, here, um, we have the project. This is a typical project setting for BLO photo. On the left hand side we have the project configuration and we have, uh, some sample images on the right hand side so that you can easily see how the project settings will reflect, uh, on your output images. So you can remove the background, set the background as transparent, any other solid color, or you can also choose any background, any custom image to be applied as the background of your image. So in the next step, we are doing content a very sized, this is specifically what Rahul was talking about. So our, an AI engine has the capability of identifying the object inside the image and then resizing based on that object.  
Joyita       00:12:56    So here you can see that the object is in a, the image is in a landscape mode and the VA is aligned towards the left of the image. But now let's say if we want to, uh, resize it to a portrait mode, uh, our AI engine is capable of, um, de sizing based on the object and keeping the object in focus and cropping out the irrelevant facts of the image. Uh, so this, uh, it doesn't matter whether the object is aligned towards a corner, so the top bottom right, uh, AI engine will always recognize where the object is placed and crop the image accordingly. Now, the, uh, second image image, second feature of cloud do photo, which is very unique to cloud, cloud do photo, is, uh, preserving the site contact. So here you can see that we have, uh, specified a margin of 8% to be applied to all sites, but our AI engine is capable of recognizing that the object is touching the edge of the image and it'll not add a margin to that size, so site so that, um, the image doesn't look truncated.  
Joyita       00:14:02    So in the next step, uh, you can add a drop shadow, like Raul was mentioning, uh, through, uh, you can add, uh, the similar kind of shadow to all your images. You can, uh, control the size of the shadow using the vertical and horizontal settings. You can also change the blur intensity and the color of the shadow. So in the next step, you can select all the output sizes that you want to, uh, generate for your images. So you may have different, uh, requirements for different marketplaces. So you can select all these sizes at one go. And, uh, when you run the project once, you will get all the, uh, all the output sizes, uh, in one run of your project.  
Rahul        00:14:42    I think joy, that that's probably the most important point that, uh, what you just showed there for different output sizes. Yeah, uh, because a lot of, uh, our retouching friends, they need to target their own website, various marketing, uh, channels, uh, there's, uh, uh, CMSs. And so there's often need to crop different ways aspect ratios have to be changed. And being able to first set up your preferences on what are the target sizes and what, what does the retouching need to to do, uh, you have the ability then that that's what you're going to show next to then take your own images and apply it to, uh, all of them the same settings and generate multiple crops.  
Joyita       00:15:23    Right, right. You're absolutely right, right, Raul. And because of the content of resizing, um, our, uh, uh, basically AI engine will be able to resize them. Uh, you can give any output sizes, it will be able to resize them, keeping the object in focus as well as maintaining the aspect ratio. So here I just uploaded 10 images of, uh, my choice. And the good part about this is once you run the, uh, project, all the, uh, images will be cropped and resized as per your settings at one go. And I have only uploaded 10 images. You can upload, uh, thousands of hundreds of images in this, uh, in this page. And another good part is that you don't need to set up the project again and again. Uh, if you want to run the same configuration on a different set of images, you can just create a duplicate of this project. All the configuration will be cro copied over and you can just upload a different batch of images, um, to your pro, uh, to the, uh, second project. And then you can just, uh, run the same, um, automation on that, uh, different set of, uh, images as well.  
Joyita       00:16:29    So, uh, I have already basically run this automation on the same, um, images and I will just show you how the output looks like. So if you look at the outputs, you can see that all the images have been resized keeping the object in focus. So I will show a couple of examples. Like for example, in this shoe, the, um, in this image, the shoe was aligned towards the right of the image. But here you can see that even though we have cropped the image to a square dimension, uh, the shoes are still in focus. Uh, in this, in the case of this image, the shoe was aligned towards the bottom of the image, but here we have been able to again, resize the image, keeping the shoe in focus in this, uh, image. The tools were towards the bottom left corner. And uh, we have, uh, resized the image keeping the tools in focus and also this also demonstrate the preserving of the edge contact.  
Joyita       00:17:29    That is the tools for touching the edge of the image. That's why we have not added any margin to this site. So now, um, if we want to compare this with, uh, Photoshop batch actions, I had actually run the same, uh, Photoshop batch action on the same set of images. So I had run, uh, removed BG batch action. So here you can see that for some of the images, the remove background using Photoshop was not very accurate. And uh, this will definitely need some manual effort to get the correct result, whereas in our case, the images, but the background removal worked perfectly fine  
Rahul        00:18:09    And enjoy that. That's probably, sorry to interrupt, but that's yeah, a very key point 'cause I know a lot of retouchers and retouching firms are used to running Photoshop actions. And this question often comes up that can AI do better than the batch actions that we can perform with, uh, Photoshop? And I think you're answering that.  
Joyita       00:18:28    Yeah, exactly. So here, basically without any manual intervention, you can see that, uh, how nicely the background has been removed and also the drop shadow has been added. So, uh, another thing, uh, uh, speaking of the resize that we are doing, uh, so I also, uh, try to use Photoshop batch action for resizing the image using the same dimension that I was using in BLO photo. So here, basically, um, you can see that, uh, the images, since the images were of different dimensions and also the object alignment varied across the images, uh, Photoshop, because it doesn't have the AI capability of identifying the objects, it was not able to, um, detect the objects. So Photoshop batch action, basically the same action will not work if the image dimensions are not uniform and the object alignments are varying across, uh, across images. Whereas in the case of product photo, uh, because it has the AI capability of recognizing the object position, it is able to detect where the object is and resizing it, uh, based on the position of the object.  
Rahul        00:19:35    Yeah, that's another great point because there are a lot of tools that have bulk or batch action, but being able to combine the intelligence of various AI models, and there are not one but multiple AI models that we are running in order to smartly resize and crop these images as how a e-commerce seller would want them them to be.  
Joyita       00:20:00    Yeah. Right. Uh, yeah, that is what actually I wanted to highlight also, uh, in the live demo. Uh, so one another, uh, the second use case I would like to quickly show is, uh, Rahul, like you mentioned in the shopping website, uh, generating unrecognizable crops for models. So we can do that also using crop do photo. So basically here I had set a crop reference point between eyes and nose, and here you can see that the, uh, images of the models have been cropped from that reference point. So one specific case I would like to point out here is, uh, for this model, uh, even when the model is not facing the camera, we, uh, the r AI engine is still able to recognize the reference point and crop from that reference point. So, uh, the position of the model doesn't really matter. Our AI engine is smart enough to recognize the reference point, even in this case.  
Joyita       00:20:55    And even in this case, you can, uh, generate multiple output sizes. Uh, so yeah, it's similar to the background removal, uh, type of logic. And, uh, the last use case that I would like to highlight is, uh, like in the previous project, I, uh, cropped from a reference point, uh, you can also crop around a reference point. So this may be specifically useful for, let's say, generating ID cards for your business. So here I had, um, uh, configure the project to crop around a reference point and remove the background and, uh, set a custom color so you can, uh, set a standard size that is your company's requirement or any custom color as the background, as per your company requirement. And then you can crop images around a particular reference point. So yeah, these are the, some of the use cases. I wanted to show how AI can, uh, how cloud.photo uses his AI capabilities to help editing images in <inaudible>.  
Rahul        00:21:49    That's great. Uh, uh, jota. So I think one point that I would like to make is, uh, jota is going through a couple of, uh, use cases here, but AI here is much more powerful. So sometimes as, uh, editor or re toucher, uh, it's kind of hard to visualize how to get the AI to do what you have in mind. It can do a lot of things, but it cannot read your mind. So you can play with the settings, but we always encourage our users to chat with us. If you are trying to manage a particular type of crop and you don't think you understand how to get the exact results, just hit the chat button in our app and our AI expert can, can help tune the AI to, to specifically what you need. Alright, so Jota, I'm gonna take the control o back from you if that's okay.  
Joyita       00:22:36    Yeah, sure. Go ahead Raul. Thank you.  
Rahul        00:22:39    Okay, just gimme a second here. Okay, so back to the presentation here. So what Joy, uh, went through was existing capabilities of AI for some basic tasks, cropping, resizing, content aware object detection, but it doesn't end there. We know as uh, e-commerce sellers, you have a lot of high end tasks that you need to, to deal with. And so we are working actively on adding more and more AI capabilities for various kind of images. So one of the most requested feature is being able to handle lifestyle images even better. So what I mean by that is if you look at, for example, this card, say you are trying to sell the, the, uh, umbrella here as opposed to the chairs. So what we would be able to do very soon is able to detect various kind of products in our image. There might be three or four products that a I sees, but you are interested only in beach umbrellas.  
Rahul        00:23:46    So you would be able to set a theme that, hey, I'm trying to crop a batch of images and it's all going to be about hand towels or beach umbrellas or whatever product. And our AI would basically then be able to zoom into the specific, uh, uh, section of the image, create closeups, and create, uh, tight fits basically with the content aware resizing. So that's something that we are working on and it'll be released very soon. Another thing that comes up a lot is lot of sellers have lot of product skews that they need to shoot for. And as we know, photoshoot can get expensive. And, um, what we are looking at doing is using generative AI models for automatically filling back the, the backgrounds. So for example, if you had, uh, a power tool like this, the background may need to be like a shop, uh, where the, the tool basically makes sense.  
Rahul        00:24:41    So using a chat g p t like chat interface soon, you would be able to describe the backgrounds that you want and we will be able to fill them and put your object on, on top. And the challenge here is that, again, there are a lot of generative AI model. There is tremendous amount of buzz in the media about generative ai, but it's very hard to control them by what? By, by that, what I mean is as an e-commerce shop, you have very specific requirements. You can't have, for example, this red color change to green because the generative AI model figured out that's the best way it'll contrast with the background. So that's where e-commerce optimize AI models come in, and that's a huge part of what we are working on next. And there are a number of other features that you will see soon.  
Rahul        00:25:26    Image upscaling, being able to clean blemishes, being able to rename, uh, files. And it's, it's kind of silly that in 2023 we are talking about re automation, but a lot of customers have given us feedback that that's a huge problem. And why is that? The reason is very simple. When you get images from photo shoots or other vendors give you images, they often have funny names. Image might be called image 1 2 8 4 dot p n g, but you want to have your final names be logical, you want to put your product queue, your style name in the file. And right now you may spend a lot of time doing that by hand. You might be, if you're a larger brand, you might be using a photo studio operations system to go and manage the metadata and have the system automatically renamed the file. But that's a lot of tech to achieve something as simple as a filing and using AI soon it'll be able to recognize that this batch of images is really the same product and be able to rename the, the files.  
Rahul        00:26:26    So we are almost at the end of the webinar, there's lot to talk about and show. Uh, so I would encourage you to interact with us through, through our website, through our chat. But to, to conclude what we showed you today is that AI models that are optimized for e-commerce and I I emphasize optimize for e-commerce because there are a lot of AI models out there. They can currently handle low to mid end retouching needs at, at scale. You would have no problem handling a hundred images or a thousand images, uh, in, in, in one batch. Now key to success as we see it is how do you split your workflow between AI versus manual? 'cause like we have been saying, there are a lot of tasks that will still require manual retouching. You might have, um, need to fix the skin tones or do other high-end operations.  
Rahul        00:27:20    So for sure manual is not going away if Photoshop is not going away. So if you can take a batch of images and say, Hey, vast majority of these we can pass through AI and we will be able to do these routine tasks through AI really fast and cheaply, then you have a win. But if you find that no, I need every single image through go to go through a manual retouching flow, then you might find the R o i or return on investment from AI is, is just not there. But what we find in practice is most of the retailers, most of the, uh, e-commerce seller are easily able to, uh, figure out, uh, a batching strategy and send some stuff to ai like our tool and, and some stuff to, to manual and then AI tool, like crop photo even have options for manually editing after AI is done.  
Rahul        00:28:08    And I cannot emphasize that enough. AI is great, but it's not perfect. So at two times you can jump into a manual mode. So by the way of looking at, uh, at pricing and cost, I would encourage you to take a look at our site. If you go over to the pricing site, we have built a, a very simple calculator that helps you determine if AI will be really cost effective for you or not. So all you have to do is enter your cadence, like how frequently do you retouch images, how many images do you retouch, say 200, the operations that you need to do on the images and the amount of time and roughly the cost per image. If you're using your own uh, retouchers or if you're using a third party, you can go and update all of that. And if you just click on calculate savings, it'll be able to show you whether AI is a possibility for you or, or not.  
Rahul        00:29:00    So yeah, just go ahead and try out this, uh, amazing calculator and check the savings for yourself. So overall, we see if companies are able to figure out how to split their workflow and take advantage of ai, they can see their retouching teams productivity go up by two to three times and retouching costs come down significantly like order of magnitude three or four times. Now this is all about, uh, this is all you can achieve today with what is there, but if you want to track where we are next, I'll encourage you to go to our roadmap on our website and you can see how AI is evolving. So that brings us to the q and a part of this, uh, uh, uh, webinar. So if you have questions then we'll be happy to, uh, take them. Alright, so the first question I see here is what's the general turnaround time using your tool?  
Rahul        00:29:54    So the answer to that is, um, it depends on the batch of assets that you are working with. So example, if you have a batch of 10 or 20 assets, then you can expect turnaround times to be like a minute, minute and a half. If you have a batch of say a hundred images, then it might take three or uh, four minutes. And, uh, compared to what it takes manually, that, that's a huge amount of time saved. Okay, let's see what other questions are, are there, so somebody is asking, we do over 2000 images a month for our wholesale channels, and these are mostly on model imagery. Do you have customers who have handled such high volume? And and the answer to that is ab absolutely. We have major fashion brands that are using crop.photo today to do exactly that. In fact, one of our customer in east coast is doing around 3000 images every month and they do these, uh, unrecognizable headless crops that they need to do before they upload to likes of Macy's and Norstrom. So yeah, absolutely you can, can do that. Hey, another question is, is AI suitable for all image types? Are there any limitations where we cannot use certain, um, types of images with ai? So maybe Joya you can answer that question.  
Joyita       00:31:20    Yeah, so currently, like you saw, um, the AI is uh, suitable for, uh, uh, all the object types, but uh, right now we are supporting only jpeg P N G and WebP type of images. But, uh, we are like Raul mentioned in the roadmap that we will be soon, uh, supporting content sizing for lifestyle images. So that is in our roadmap and we are actively working on the specific requirement to release it as soon as possible.  
Rahul        00:31:48    Thanks, jota. Another question that is popping up, uh, here Jota that you can answer is, how is this AI better than other tools such as Photoshop where we can use batch actions to do bulk editing? And I know you talked about, uh, batch actions, but maybe you can, uh, just quickly, uh, mention this point about batch editing with other tools like Photoshop versus our AI tool,  
Joyita       00:32:13    Right? Yeah. So like you saw, um, in case of batch action, it becomes very tricky when the image dimensions are different and when the object alignments are different, it becomes very tricky because Photoshop doesn't have the capability AI capability of identifying the position of object. But, uh, because scrub.photo has the AI engine which identifies the position of the object, uh, container size works much better when the image dimensions are varying and when the object alignment is varying in the image. And another, uh, another good thing is like, um, all this processing is happening on cloud. In case of Photoshop, it'll use up your system resources, but here, uh, since uh, it's happening on cloud, none of your system resources are uh, affected. So you don't have to have any specific system specification for, um, a running cloud do photo.  
Rahul        00:33:03    Okay, great. So another question that's coming up is how many images can it handle at once? So the good news is that our AI technology runs truly in the cloud because there are a lot of AI photo editors. If you Google, you will come across many. And one of the thing that you will see is they often run in your browser, which really means they're running on your desktop. So if you were to give them a batch of 2050 assets, they, they choke. And with the crop.photo, once you upload your images, like Joya was saying, uh, it all runs in the cloud. So there is really no limit You can hand us, uh, a hundred images, thousand images, 10,000 images, we can handle it at all.  
Rahul        00:33:49    Hey, uh, let's see, another question. I have a va a virtual assistant, uh, that is what VA stands for, uh, I use with an online AI photo editing app. How is rock.photo better or more cost effective? That's a great question because uh, a lot of you, uh, especially the small business owners, Shopify store owners have the need to optimize cost and you often find the, the cheapest way of, uh, uh, getting a task done. And sometimes you use VA's, uh, offshore. So why, why is crop photo able to better them? The first answer is, uh, uh, consistency. So with the AI tool that's optimized for e-commerce, like you saw in Joy's demo, when you're removing backgrounds or doing crops, it's, it's basically able to just do that day in, day out consistently. That's one. And then secondly, the cost per image is, is pretty small. So you can start as low as like sense per images and it's hard for uh, any, uh, VA to, to better that. Uh, of course the high-end tasks you will still want your retouching firm or VA to be involved with. But in terms of the bulk of, uh, retouching, which is low to mid end, we can do that quicker and, and uh, faster and more reliably.  
Rahul        00:35:12    Another question is, uh, can I keep my images that I have uploaded into your cloud app and reuse them for different crops over a period of two to three months? Joya, do you wanna answer that?  
Joyita       00:35:29    Yeah, sure, Rahul. So yes, if you are subscribed to a paid plan for cloud.photo, you can reuse your images. We will be saving the images in cloud for future use. And then, like I mentioned about duplication of projects, you can duplicate a project and then, uh, if you want to run a a different project on the same set of image, you can duplicate the project, change the project settings, and then run the project on um, your existing images.  
Rahul        00:35:55    Okay. Okay. Uh, another question is what if the image is jpeg? Uh, but I want to output the image in a p and g format. Does this tool handle such operations? I can answer that. So right now, the tool is able to basically take the input format that you have, jpeg, P n g and other formats and output it to the same format unless you are doing transparency. So if for example, you remove background from a JPEG image and you want to save the transparency, then it'll automatically convert that to P N G. So it's smart about the output file, but if you want to manually control the output, uh, file types and, and um, formats, then very soon in the next uh, few weeks we'll be releasing an update that will allow you to do that precisely. So I think those are, that's all the time we have for, uh, q and a.  
Rahul        00:36:51    So like I promised we have some freebies and what we are going to offer is anybody who has attended this webinar, if you were to go ahead and book a meeting with us through this, uh, Calendly tool and our team is going to send this in the chat and we'll also email this. So if you could book a meeting with us and tell us about your challenges with, uh, retouching, we will be able to validate your webinar attendance and we would be able to then offer you a promo code for a month three on our solo plan. So with that, we come to the end of this webinar, webinar. I want to thank you for attending. We'll be sending a recording of this to all the attendees and if you could not attend, uh, you will be able to access the recording. And so take care. Bye.  
Joyita       00:37:42    Thank you.