Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. template coverage will be largely incomplete or when templates may be contaminated by transients present at the time of acquisition. We present multiband light curves for a sample of known transients, along with new transient candidates identified through our search. Finally, we discuss the prospects of applying this pipeline during the main LSST survey. Our pipeline is broadly applicable and will support studies of all transients with slowly evolving phases.

Unified Astronomy Thesaurus concepts: Core-collapse supernovae (304); Supernovae (1668); Transient detection (1957) Materials only available in the online version of record: data behind figures

The wide field of view and exceptional depth of the Vera C. Rubin Observatory will usher in a new era for time-domain astronomy (LSST Science Collaboration et al. 2009). Expected to begin full operations in late 2025, the Rubin Observatory's decade-long Legacy Survey of Space and Time (LSST; Ž. Ivezić et al. 2019) LSST's nightly alerts rely on difference imaging, which compares new observations to deep reference templates to identify brightness fluctuations from variable and transient sources. Template collection is expected to continue through the first year of regular survey operations (L. P. Guy et al. 2023Guy et al. , 2025)). Transients observed during this early period may contaminate templates, making it difficult to accurately distinguish their flux from contaminating background light (e.g., from the host galaxy). Such contamination may not significantly affect the transient detections for rapidly evolving transients or supernovae (SNe) that change brightness on relatively short timescales. Despite inaccurate reported photometry, these transients can still be detected by the LSST alert stream even if partially imprinted in the templates.

In contrast, long-duration transients that do not exhibit strong luminosity evolution will be difficult to identify in real time if they are present in the LSST templates. This includes long-lived precursors to core-collapse SNe (CCSNe; A. Pastorello et al. 2007;N. L. Strotjohann et al. 2021;D. Tsuna et al. 2023D. Tsuna et al. , 2024a; S. J. Brennan et al. 2025) such as those detected in SN 2020tlf (W. V. Jacobson-Galán et al. 2022), SN 2023fyq (S. J. Brennan et al. 2024;Y. Dong et al. 2024), SN 2023zkd (A. Gagliano et al. 2025), as well as longduration superluminous SNe (e.g., S. Gomez et al. 2024) and luminous red novae (LRNe; J. C. Mauerhan et al. 2015;N. Smith et al. 2016;N. Blagorodnova et al. 2017). Even if such transients are detected, obtaining clean photometry will require waiting until they have faded, at which point templates can be constructed. This delay could hinder timely follow-up and downstream analysis during the early phases of the LSST survey.

The release of Rubin DP1 provides a valuable dataset for developing and testing infrastructure that can be applied to real LSST data. In this Letter, we present the SLIDE package for performing LSST image subtraction using images taken by the Dark Energy Camera (DECam; K. Honscheid & D. L. DePoy 2008;B. Flaugher et al. 2015). This provides an alternative to LSST-derived templates and will benefit the broader transient community in the initial years of the LSST survey.

We have released our SLIDE package on GitHub. 35 SLIDE is intended to be run directly on the Rubin Science Platform (G. Dubois-Felsmann et al. 2019;W. O'Mullane et al. 2024). We use SLIDE to search for DP1 transients within two of the extragalactic fields observed by LSSTComCam: the Extended Chandra Deep Field South (ECDFS) and the Euclid Deep Field South (EDFS). We find that all transients reported to the Transient Name Server (TNS; A. Gal-Yam 2021) within our selected search area are successfully recovered, provided they had not already faded by the time the images were taken. In addition, we identify 29 previously unreported transients, 18 of which are likely nuclear transients, and 12 of which are either not present or have fewer than two detections in DP1's DiaObject catalog (NSF-DOE Vera C. Rubin Observatory 2025c).

In Section 2, we provide an overview of SLIDE and test it on a Rubin LSSTComCam transient with template contamination. In Section 3, we describe our search for transient candidates in the EDFS and ECDFS fields using our corrected difference images. We summarize our findings and outline future prospects with the full LSST data stream in Section 4.

SLIDE can be easily installed on the Rubin Science Platform 36,37 (W. O'Mullane et al. 2024). We include an example notebook to demonstrate its usage. 38 Here, we outline its major components.

DECam is a wide-field charge-coupled device (CCD) imager mounted on the 4 m Blanco telescope at Cerro Tololo Inter-American Observatory (CTIO) in northern Chile. Initially designed for the Dark Energy Survey (DES; Dark Energy Survey Collaboration et al. 2016), DECam consists of 62 science CCDs with a pixel scale of 0.263 pixel -1 and a field of view of approximately 3 deg 2 . The DES survey was conducted from 2013 August 15 to 2019 January 9 and covered 5000 deg 2 in grizY bands. SLIDE automatically retrieves deep coadded images from the DES Data Release 2 (DR2; T. M. C. Abbott et al. 2021) that overlap a position of interest. The final coadds reach a median 5σ depth of g = 25.4, r = 25.1, i = 24.5, z = 23.8, and Y = 22.4, which are deeper than single-visit depths expected from LSST and deeper than images released as part of DP1 (F. B. Bianco et al. 2022; NSF-DOE Vera C. Rubin Observatory 2025b). This makes DES DR2 images suitable as templates for LSST image subtraction. All the difference images used in this Letter are made using the DES DR2 templates.

Alternatively, SLIDE can retrieve coadded DECam images from the Dark Energy Camera Legacy Survey (DECaLS; R. D. Blum et al. 2016;A. Dey et al. 2019) for use as templates. These images have similar or slightly greater depth than single exposures in DP1, making them a useful alternative when DES templates are not available. Users may also supply custom DECam templates.

DES DR2 (or DECaLS) images are retrieved using the Simple Image Access service provided by the Astro Data Lab (M. J. Fitzpatrick et al. 2014;R. Nikutta et al. 2020). DECam images are aligned and rescaled to match the LSST images using the reproject package,foot_6 which uses an adaptive, antialiased resampling algorithm (C. E. DeForest 2004). The point-spread function (PSF) of the DECam images is modeled using Photutils with stars selected from the Gaia DR3 catalog (Gaia Collaboration et al. 2016, 2021;M. Riello et al. 2021).

For DP1 images, we obtain the calibrated exposure images (visit_image) from the Rubin Science Platform using the Butler (T. Jenness et al. 2022). These images have been processed by the LSST Science Pipelines (Rubin Observatory Science Pipelines Developers 2025) and are ready for scientific use. LSSTComCam consists of nine CCDs; SLIDE can operate on either full CCD images or cutout regions. The LSSTComCam image PSFs provided can vary slightly over the field of view. Therefore, we use the median PSF of the detector for image subtraction. Alternatively, our package offers options to recalculate the PSF and refine the World Coordinate System of images using stars from Gaia DR3.

The image subtraction is performed using a Python implementation of the Zackay-Ofek-Gal-Yam (ZOGY; B. Zackay et al. 2016;D. Guevel & G. Hosseinzadeh 2017) algorithm,foot_7 which provides mathematically optimal statistics for image subtraction and does not require that the reference image has a sharper PSF than the science image. The runtime for PSF construction and image subtraction depends on image size: for a full LSST CCD image (4000 × 4000 pixels), it takes approximately 1.5 minutes, while for a 1500 × 1500 pixel cutout, it takes about 15 s.

Finally, PSF photometry is performed on the difference images at specified positions (R.A. and decl.) using the Photutils package from astropy (Astropy Collaboration et al. 2022).

We test SLIDE on a known transient, AT 2024ahzi, which was reported to TNS on 2025 March 13 (I. Andreoni et al. 2025; C. T. Murphey et al. 2025). Flux from AT 2024ahzi is present in the LSSTComCam templates used by DP1, making the reported difference-imaged forced photometry unreliable (K. de Soto 2025, in preparation). We process all available DES images overlapping the transient position and find that the resulting photometry is consistent with that obtained by the Young Supernova Experiment (D. O. Jones et al. 2021;P. D. Aleo et al. 2023) using DECam within ∼1σ (see detailed photometric comparison in K. de Soto 2025, in preparation). In Figure 1, we show examples of image subtraction at the position of AT 2024ahzi using the DES templates as a demonstration of the subtraction quality. The subtractions are generally clean, and the transient is clearly detected in the center when present.

Rubin DP1 contains ugrizy images from seven fields, taken with LSSTComCam between 2024 November and December. We select the ECDFS and the EDFS fields to perform an experimental transient search, as these are the most wellobserved fields in DP1, are far from the Galactic plane, and have sufficiently overlapping coverage by DES. For each field, we select a subset of r-band exposures that maximizes overall spatial overlap across visits while also ensuring even temporal coverage. We restrict the transient search to a single filter to reduce the computational workload; we prefer the r band as it has the highest cadence in both fields. This selection yields 37 visits of ECDFS and 18 visits of EDFS. For each visit, image subtraction is performed on each CCD's image independently.

Transient detection is performed on the difference images using SEP (K. Barbary 2016), a Python implementation of Source Extractor (E. Bertin & S. Arnouts 1996).

Stars brighter than approximately 16 mag saturate LSSTComCam's 30 s exposures. Although these bright stars are masked out in the difference images, we find that they often produce prominent spike-like artifacts in the surrounding area, which can be misidentified as transient detections. To mitigate contamination, we exclude any candidates located within 20″ of such stars. Additionally, we find that stars with proper motion may be misaligned between the template and science images. Such misalignment can create residual artifacts in the difference images, and we therefore exclude candidates located within one full width at half-maximum of sources classified as stars in the DES DR2 catalog (T. M. C. Abbott et al. 2021).

To identify transients of interest from the remaining 15,785 targets, we require that each candidate has at least three detections, which removes contamination from cosmic rays and other artifacts. We also require that the peak-to-peak flux variation exceed 3 times the mean flux uncertainty, that the standard deviation of the flux exceed the mean flux uncertainty, and that the peak-to-peak magnitude variation be greater than 0.3 mag. Sources that do not meet all of these criteria are excluded from further analysis.

We associate the remaining 1224 candidates with likely host galaxies using Pröstfoot_8 (A. Gagliano et al. 2025). Pröst calculates the posterior probability that each galaxy in a given search region is the true host galaxy using the fractional offset, redshift, and brightness of the host/transient. We use a search radius of 60" and consider galaxies in the Galaxy List for Advanced Detector Era catalog (GLADE; G. Dálya et al. 2022), Panoramic Survey Telescope and Rapid Response System (Pan-STARRS; K. C. Chambers et al. 2016) Data Release 2 (DR2; H. A. Flewelling et al. 2020), and the DeCaLS Data Release 10. We also flag nuclear transients using iinuclearfoot_9 (S. Gomez 2024), which determines whether the location of the transient coincides with the center of its host galaxy with sufficient probability. In our final candidate selection, we perform human vetting to select the most promising candidates and retain only transients, both nuclear and non-nuclear, with confident host associations.

Our criteria yield 39 transient candidates: 22 in ECDFS and 17 in EDFS. The detections for these candidates are separated by at least 20 days, which effectively excludes moving objects. These candidates are listed in Tables 1 and 2, respectively.

3.3.1. Known Transients I. Andreoni et al. (2025) reported three newly identified extragalactic transient candidates and eight previously reported transients as visible in the DP1 images. These transients were required to have confident host galaxy associations and not to be colocated with star-like objects or galactic nuclei. Among the seven reported transients in the ECDFS/EDFS Murphey et al. 2025) lie within the field we selected but are not detected, as they have fewer than three detections in our selected visits. We note that the photometric classifications of these TNS transients have been discussed in J. Freeburn et al. (2025d), and refer the reader to that work for further details. We perform PSF photometry using SLIDE at the positions of AT 2024ahyy, AT 2024ahzc, AT 2024aigs, AT 2024aigg, AT 2024aigl, AT 2024aigj, AT 2024aigk, AT 2024aigh, AT 2024aigw, AT 2024ahwk, AT 2024aigt, AT 2024ahsx, AT 2024ahyq, and AT 2024aigv. We refer the reader to K. de Soto (2025, in preparation) for photometric analysis of AT 2024ahzi. Our transient search uses a subset of r-band images, which may not always optimally cover individual objects. To obtain multiband and better temporal coverage for specific objects, we search all DP1 images that overlap with each object's position and select up to two images per night per filter (to reduce computational cost). We then generated difference images and extracted photometry from 1500 × 1500 pixel cutouts centered on each object using SLIDE across all filters. Since there are no u-band observations in DES and few y-band observations in DP1, we performed image subtraction only on the griz-band images. The light curves and host properties of these objects are shown in Figure 2.

We cross-match the remaining 29 candidates with the DiaObject table on the Rubin Science Platform, considering only DiaObjects with at least two detections. A total of 17 transients have a corresponding DiaObject within 1″. The remaining 12 transients are either missing or have less than two detections in the DP1 DIA catalog, because transient flux is contaminating the templates and pushing the difference flux variation below the detection threshold. This highlights the importance of robust templates in transient identification. We show examples of transients with and without DiaObject object associations in Figure 3.

In this Letter, we do not perform detailed light-curve analysis on our identified candidates; however, as a demonstration of the future workflow in the LSST era, we derive the host properties of each transient using FrankenBlast 1 and 2, as well as in Figure 2. Host properties, especially photometric redshifts, can significantly improve the accuracy of photometric classifiers, particularly at early times (K. Boone 2019; D. Muthukrishna et al. 2019;S. Gomez et al. 2020;P. Sánchez-Sáez et al. 2021;A. Gagliano et al. 2023;M. Kisley et al. 2023;K. M. de Soto et al. 2024;X. Sheng et al. 2024;V. A. Villar et al. 2024;A. Boesky et al. 2025;R. Gupta et al. 2025).

We present the SLIDE package, which performs LSST image subtraction using DECam templates. To demonstrate the anticipated workflow for the LSST survey, we conduct an experimental transient search via the Rubin Science Platform using DP1 difference images produced by this package. We present multiband photometry and host galaxy properties for the most promising transients.

This Letter demonstrates the potential of SLIDE to uncover extended preexplosion or long-duration transient activity within LSST. In recent years, it has been found that many CCSNe, such as normal SNe II, SNe IIn, and SNe Ibn, interact with dense circumstellar material (CSM) around their progenitors (D. C. Leonard et al. 2000;N. Smith et al. 2015;D. Khazov et al. 2016;O. Yaron et al. 2017;A. Gangopadhyay et al. 2020A. Gangopadhyay et al. , 2025;;V. Morozova et al. 2020; S.-Q. Wang & L. Li 2020; R. J. Bruch et al. 2021Bruch et al. , 2023;;X. Wang et al. 2021;C. Pellegrino et al. 2022;G. Terreran et al. 2022;T. Ben-Ami et al. 2023;K. A. Bostroem et al. 2023aK. A. Bostroem et al. , 2023b;;G. Hosseinzadeh et al. 2023;J. Pearson et al. 2023;J. E. Andrews et al. 2024J. E. Andrews et al. , 2025;;Y. Dong et al. 2024;W. V. Jacobson-Galán et al. 2024;N. Meza-Retamal et al. 2024;M. Shrestha et al. 2024aM. Shrestha et al. , 2024b;;S. J. Brennan et al. 2025; Z. Y. Wang et al. 2025), likely produced months to years before their final explosions. The origin of this CSM, its geometry, and its implications for CCSN progenitor systems remain hotly debated (R. A.

Chevalier 2012; E. Quataert & J. Shiode 2012; N. Soker 2013; J. H. Shiode & E. Quataert 2014; N. Smith & W. D. Arnett 2014; J. Fuller 2017; V. Morozova et al. 2020; L. Dessart et al. 2022; B. D. Metzger 2022; S. C. Wu & J. Fuller 2022; D. Tsuna et al. 2024aTsuna et al. , 2024b)).

Improved characterization of precursor emission provides critical insights into the final stages of stellar evolution, and may serve as an early warning signal for imminent CCSNe (D. Tsuna et al. 2023). To date, precursor activity has been most commonly observed in SNe IIn (N. L. Strotjohann et al. 2021;D. Farias et al. 2024; S. J. Brennan et al. 2025;A. Gagliano et al. 2025;A. Pastorello et al. 2025). In contrast, only three SNe Ibn (A. Pastorello et al. 2007;N. L. Strotjohann et al. 2021; S. J. Brennan et al. 2024;Y. Dong et al. 2024) and a single SN II (W. V. Jacobson-Galán et al. 2022) have shown evidence for precursor emission, potentially due to their fainter intrinsic luminosities. Moreover, precursor spectroscopy, critical for probing progenitor systems, has only been published for a few events (A. Pastorello et al. 2013; S. J. Brennan et al. 2024). Building a larger sample of SNe with detected precursor emission and precursor spectra is essential for constraining their occurrence rates and understanding their underlying physical mechanisms.

Furthermore, LSST is expected to drastically increase the number of photometrically identified, long-duration precursor events. With a single LSST visit, precursors from normal SNe II and SNe IIn can be detected to approximately 100 and 300 Mpc, respectively (W. V. Jacobson-Galán et al. 2022;A. Gagliano et al. 2025), while precursors of SNe Ibn can be detected to approximately 150 Mpc (e.g., Y. Dong et al. 2024). Therefore, precursor searches can be constrained to nearby, bright, low-extinction galaxies to reduce computational load and enable rapid identification and follow-up of promising candidates. Host association (e.g., with algorithms such as Pröst as we described in Section 3.2) is essential for targeted searches to further decrease the computational load on the Rubin Science Platform. Host galaxy properties can be derived using Blast and FrankenBlast, further enabling prioritization and classification. We expect SLIDE, paired with the approach outlined in this Letter, to be particularly effective in discovering these long-duration precursors.

Our package will support the study of all transients with slowly evolving phases. LRNe (J. C. Mauerhan et al. 2015;N. Smith et al. 2016;N. Blagorodnova et al. 2017), for example, are generally understood to be the product of common envelope episodes and, potentially, mergers (e.g., N. Soker & R. Tylenda 2003;R. Tylenda et al. 2011;B. D. Metzger & O. Pejcha 2017;N. Soker 2024). LRNe often undergo gradual brightening that lasts for years prior to the main outburst. Multiband photometric and spectroscopic observations during the preoutburst brightening phase can 1 and 2. (The data used to create this figure are available in the online article.)

Table 1 ECDFS Transient Candidates ID R.A. Decl. Nuc. Flag z host DIA Object ID ( ) / Z Z log ( ) / * M M log Age SFR (hh:mm:ss) (dd:mm:ss) (Gyr) (M ⊙ /yr) Transients Reported to TNS 2024ahsx 03:33:28.07 -28:12:54.36 1 0.261(0.011) 611253629533291776 + 0.26 0.32 0.26 + 10.87 0.25 0.62 + 0.59 0.57 5.98 + 825.08 600.86 493.09 2024ahwk 03:29:50.944 -28:13:04.73 0 0.270(0.013) 611253973130674268 ⋯ ⋯ ⋯ ⋯ 2024ahyq 03:31:37.65 -28:20:01.31 1 0.294(0.040) 609782139377943168 + 0.50 0.53 0.30 + 10.75 0.10 0.10 + 5.54 2.08 1.54 + 3.00 2.11 6.84 2024ahyy 03:31:34.22 -28:24:45.37 0 0.438(0.105) 609781520902651904 + 0.80 0.86 0.54 + 9.67 0.65 0.28 + 3.54 3.31 1.57 + 8.24 4.64 7.07 2024ahzc 03:31:21.18 -28:16:47.64 0 0.290(0.042) 609782208097419264 + 0.75 0.72 0.61 + 10.38 0.13 0.12 + 6.85 0.92 2.73 + 1.77 1.27 6.31 2024aigg 03:32:29.94 -27:44:23.33 0 0.069(0.015) 611255759837069440 + 0.73 0.43 0.32 + 10.26 0.09 0.11 + 3.94 1.86 1.57 + 1.49 0.76 1.52 2024aigj 03:32:51.02 -27:40:52.60 0 0.251(0.047) 611256447031836800 ⋯ ⋯ ⋯ ⋯ 2024aigt 03:33:41.37 -28:13:24.81 0 0.296(0.053) 611253629533290624 ⋯ ⋯ ⋯ ⋯ 2024aigw 03:30:55.57 -27:51:58.87 0 0.323(0.011) 611255210081255575 + 0.28 0.35 0.24 + 11.31 0.12 0.10 + 5.00 1.72 1.12 + 12.05 9.40 22.85 2024aigv 03:32:13.81 -28:28:14.40 0 0.375(0.055) 609788942606139423 + 0.20 0.23 0.18 + 11.17 0.05 0.05 + 5.54 0.64 0.63 + 9.74 4.03 6.06 Unreported Transients 13 03:31:37.69 -28:04:10.16 0 0.132(0.020) ⋯ + 0.91 0.47 0.55 10.10 + 0.11 0.09 5.87 + 1.66 1.11 1.51 + 0.80 2.16 14 03:31:35.11 -28:07:14.42 0 0.127(0.030) 611254522886494620 + 1.20 0.45 0.58 9.83 + 0.21 0.15 4.79 + 4.10 1.64 1.60 + 0.98 3.01 21 03:31:41.65 -28:05:10.74 1 0.205(0.056) 611254454167011721 + 0.70 0.74 0.75 9.63 + 0.68 0.26 5.55 + 5.29 1.86 3.46 + 2.61 8.91 100 03:31:34.14 -27:49:59.61 0 0.464(0.125) ⋯ -1.06 + 0.61 0.72 10.21 + 0.77 0.21 3.35 + 3.29 1.45 34.04 + 20.92 37.63 706 03:33:32.49 -27:48:32.00 1 0.333(0.108) ⋯ -0.76 + 0.88 0.66 9.64 + 0.72 0.36 3.99 + 3.87 1.82 7.05 + 4.86 9.50 1071 03:31:41.92 -28:04:22.22 0 0.149(0.033) ⋯ -0.69 + 0.79 0.63 9.41 + 0.65 0.22 4.82 + 4.79 2.42 3.54 + 2.98 11.21 1314 03:33:07.70 -27:53:31.81 1 0.121(0.036) ⋯ -0.85 + 0.65 0.71 9.21 + 0.20 0.13 6.27 + 2.76 1.42 0.32 + 0.19 1.26 1350 03:32:49.30 -27:37:57.07 1 0.131(0.066) 611256447031837444 ⋯ ⋯ ⋯ ⋯ 1367 03:33:21.09 -27:39:12.07 1 0.597(0.298) 611256378312359976 ⋯ ⋯ ⋯ ⋯ 1547 03:32:46.03 -28:22:32.21 1 0.621(0.216) 609788873886662802 -0.47 + 0.53 0.44 10.59 + 0.63 0.25 3.74 + 3.65 1.08 82.07 + 45.82 91.77 1690 03:31:20.77 -27:56:49.26 1 0.834(0.118) 611255210081255450 -0.37 + 0.81 0.35 10.39 + 0.25 0.16 3.07 + 1.48 0.70 14.58 + 5.80 8.55 1897 03:33:09.50 -27:44:06.89 1 0.117(0.032) 611255691117594961 ⋯ ⋯ ⋯ ⋯ 1965 03:31:24.22 -27:53:42.27 1 0.110(0.055) 611255210081255504 ⋯ ⋯ ⋯ ⋯ 2764 03:31:47.87 -28:17:00.77 1 1.130(0.231) ⋯ -0.16 + 0.34 0.22 11.62 + 0.10 0.10 3.08 + 1.19 0.48 6.19 + 4.86 18.46 2798 03:33:46.01 -28:20:07.43 1 0.170(0.038) ⋯ -0.75 + 0.69 0.53 9.41 + 0.32 0.19 4.97 + 3.83 1.61 1.16 + 0.77 2.44

Note: Galaxy properties are not derived for hosts with insufficient photometric data.

Table 2 EDFS Transient Candidates ID R.A. Decl. Nuc. Flag z host DIA Object ID ( ) / Z Z log ( ) / * M M log Age SFR (hh:mm:ss) (dd:mm:ss) (Gyr) (M ⊙ /yr) Transients Reported to TNS 2024aigh 03:57:17.80 -48:22:08.30 0 0.06(0.04) 592915218690999552 + 0.45 0.21 0.25 + 10.66 1.28 0.51 + 7.75 0.77 1.63 + 0.03 0.03 0.12 2024aigk 03:55:31.82 -48:27:43.71 1 0.168(0.030) 592915356129952000 + 0.17 0.64 0.27 + 10.16 0.50 0.27 + 1.83 1.62 3.76 + 3.03 2.91 13.04 2024aigl 03:59:24.16 -48:46:50.53 0 0.225(0.021) 592913706862510093 -0.75 + 0.45 0.46 10.49 + 0.12 0.09 5.26 + 2.45 1.09 4.44 + 2.54 5.94 2024aigs 03:56:53.23 -49:06:18.06 0 0.393(0.147) 591819074317582336 + 0.74 0.68 0.75 + 10.03 0.27 0.13 + 4.89 2.12 0.89 + 5.59 2.94 7.39 Unreported Transients 102 03:57:09.25 -48:47:02.21 1 0.987(0.105) 592913844301464254 -0.02 + 0.25 0.15 10.96 + 0.51 0.40 1.17 + 1.09 1.67 416.87 + 207.38 291.91 124 03:56:23.21 -48:21:48.14 1 0.637(0.076) 592915287410475027 -0.25 + 0.29 0.24 11.12 + 0.54 0.31 3.68 + 3.62 1.32 281.54 + 183.53 329.42 199 03:56:48.54 -48:19:13.44 1 0.148(0.017) 592915974605242521 ⋯ ⋯ ⋯ ⋯ 206 03:57:28.37 -48:27:49.38 0 0.156(0.040) 592915218690998834 -0.29 + 0.35 0.24 10.85 + 0.07 0.06 6.41 + 0.98 0.86 2.51 + 1.31 2.75 347 03:54:35.18 -48:43:37.65 0 0.367(0.141) 592914050459894637 -0.45 + 0.85 0.40 10.53 + 0.29 0.20 5.17 + 4.16 1.51 7.61 + 5.86 20.32 357 03:54:15.75 -48:35:30.15 0 1.112(0.202) 592914737654661956 0.07 + 0.20 0.09 10.73 + 0.14 0.26 0.04 + 0.03 1.68 842.01 + 244.37 317.67 392 03:57:16.36 -48:51:04.92 1 0.829(0.240) 592913157106705461 -1.24 + 0.34 0.45 11.29 + 0.27 0.14 0.51 + 0.25 0.64 133.94 + 75.63 204.08 490 03:55:53.49 -48:22:19.23 1 0.574(0.205) 592915356129953233 ⋯ ⋯ ⋯ ⋯ 569 03:57:27.61 -48:28:38.66 0 0.406(0.159) ⋯ ⋯ ⋯ ⋯ ⋯ 668 03:56:41.15 -48:16:57.78 0 0.171(0.056) ⋯ -0.84 + 0.72 0.63 10.34 + 0.31 0.19 5.17 + 3.18 1.61 2.13 + 1.58 4.01 1236 03:58:16.51 -48:27:33.57 1 0.399(0.166) ⋯ -0.61 + 1.07 0.45 9.68 + 0.30 0.20 5.16 + 1.45 0.89 2.56 + 1.05 1.61 2507 03:56:00.29 -48:45:22.26 0 0.119(0.059) ⋯ -0.67 + 0.86 0.60 8.99 + 0.31 0.22 5.65 + 4.96 2.05 0.14 + 0.12 0.51 4777 03:55:53.52 -49:07:29.71 1 0.147(0.025) 591819143037059212 ⋯ ⋯ ⋯ ⋯ 5575 03:54:41.14 -48:34:11.92 0 0.187(0.037) ⋯ -0.68 + 0.67 0.44 9.84 + 0.31 0.18 5.18 + 3.80 1.57 4.96 + 3.17 9.11

offer valuable information about their progenitor systems and mass transfer mechanisms preceding the transient (H. Addison et al. 2022). LSST is expected to observe ∼400-800 LRNe annually (G. Howitt et al. 2020), drastically increasing the current sample of such events. Similarly, some extremely energetic transients often evolve slowly. Superluminous SNe, in particular, are a class of massive stellar explosions with luminosities significantly higher than those of normal SNe, requiring additional power sources beyond radioactive decay (A. Gal-Yam et al. 2009;L. Chomiuk et al. 2011;R. M. Quimby et al. 2011;A. Gal-Yam 2012;D. A. Howell et al. 2013;D. A. Howell 2017;T. J. Moriya et al. 2018;S. Gomez et al. 2024). LSST is expected to discover ∼10,000 hydrogen-poor superluminous events annually, with most at high redshift (z > 1; V. A. Villar et al. 2018). Similarly, ambiguous nuclear transients are energetic transients that are found in the nuclei of their host galaxies (P. J. Pessi et al. 2025;P. Wiseman et al. 2025). These are seemingly distinct from typical active galactic nuclei (AGN) and notably more extreme than "normal" tidal disruption events, although their origin is still an open question. In both cases, the extended duration of these events, especially at high redshift, implies that they are likely to be implanted in the initial LSST templates.

Early identification of transient events is essential for spectroscopic follow-up across their evolution. SLIDE enables both early detection and reliable photometry, even when the transients are embedded in their LSST template images. Because the aforementioned transient classes are rare, the early years of LSST offer an opportunity to build statistically meaningful samples of such events that will guide strategies for follow-up in the future.